Charlie Warzel | The Atlantic

How Fake People Became Real Influencers

2026-04-10T13:00:00-04:00

Subscribe here: Apple Podcasts | Spotify | YouTube

On this week’s Galaxy Brain episode, Charlie Warzel is joined by New York Times technology reporter Tiffany Hsu to discuss the rise of AI influencers—synthetic avatars, often indistinguishable from real people, that are flooding social-media feeds to sell supplements and promote brands. Hsu unpacks her reporting on the combination of forces converging around it, including the wellness industry, a historically fertile ground for scammers. The pair discuss how the volume of synthetic content online is producing a new kind of epistemic exhaustion: a fatigue so deep that many people have simply stopped caring whether what they’re seeing is real. So is authenticity already beside the point? And is an audience’s emotional response—rather than the truth behind the image—the only currency that matters?

The following is a transcript of the episode:

Tiffany Hsu: You have to create such a ridiculous volume of content, and it all has to feel fresh. Yeah; I can totally see why someone would be tempted to just make it on a computer.

[Music]

Charlie Warzel: I’m Charlie Warzel, and this is Galaxy Brain, a show where today we’re going to talk about AI influencers. There’s this post online that I think a lot about. It’s from Zachary Galia, who is this social-media content strategist. It reads: “Every post is a battle for three seconds. Platforms keep multiplying, content is seemingly endless, and attention spans are shorter than ever. Your content needs to capture your audience’s attention immediately and hang on to it for dear life.”

Maybe that stat makes your stomach drop a bit, as it does for me. What’s unquestionable, though, is that the war for attention—which is fought primarily online across this host of algorithmic infinite-scroll platforms—is being fought at almost inhuman speeds. Obsessive content marketers are in the volume game. Brands and influencers have adopted this buckshot-style approach to hawking their wares and attracting eyeballs.

And that can mean doing multiple posts about the same subject or product, but from different angles and locations—all of it to see if they can find some way to hit the sweet spot of the algorithm and go viral. It’s spamming as a strategy. And I think it’s a part of the reason why our feeds just feel so cluttered and chaotic.

And trying to feed the algorithmic beast at these inhuman speeds has meant enlisting the help of, well, not-humans. Late last year, the venture-capital firm Andreessen Horowitz invested in this company called Doublespeed, an AI company that does, quote, “bulk content creation.” In other words, it’s a bot farm. Doublespeed’s marketing is purposefully troll-y, with claims that it is, quote, “automating attention,” allowing people to create, quote, “one video a hundred ways.” The banner on the company’s website reads: Never pay a human again.

Now, it’s no secret that the internet is filling up with synthetic material or AI slop. The SEO company Graphite recently found that, beginning around November 2024, the internet experienced a slop tipping point—in which the quantity of AI-generated articles being published on the web surpassed the quantity of articles written by humans. But it’s not just text. Advances in generative-AI audio and video have meant that it has never been easier to create fake influencers from a few short text prompts. These are real-looking people. Often attractive, sometimes scantily clad women. And they’re selling real products. They’re attracting real eyeballs.

Now, some of these online influencers are pretty easy to spot, but others are good enough that they’re duping people. And in some cases, it seems almost impossible to know for certain whether a specific influencer is real or not.

We are, in essence, now just living in the uncanny valley. The stakes are real. For influencers who are worried about their jobs—but also for all of us, the people who are out here navigating this blurred reality and just trying not to get scammed or duped. So are AI influencers here to stay, or are they just this passing fad? Has the internet tipped into synthetic slop for good? How can people learn to spot what’s real and what’s fake?

To help me through all of this, I spoke with Tiffany Hsu. She’s a technology reporter at The New York Times who covers the information ecosystem, including foreign influence, political speech, and disinformation. Tiffany’s been reporting on the rise of AI influencers selling supplements and other products, and she joins me now to make sense of this very strange new world.

[Music]

Warzel: Tiffany, welcome to Galaxy Brain.

Hsu: Thanks for having me.

Warzel: So I wanted to start here. Let’s just talk very broadly. Who is Melanskia? Am I even saying that name right? Who is that person?

Hsu: Well, first of all, she’s not a person. She’s an AI avatar who looks stunningly like a real person and has apparently fooled many, many of her—think it’s now more than 300,000—followers that she is an honest-to-god human being. She’s not; she’s AI. She is meant to be an Amish lady who has several children, who posts about what you shouldn’t eat. Clean living. She talks about how she would never buy supermarket rotisserie chicken, which I saw as a personal affront. Yeah; I mean, she talks a big game about health and wellness, which is kind of surprising, given that she does not have a body.

Warzel: And that’s because she’s a generative-AI avatar, correct?

Hsu: Exactly.

Warzel: So how did you stumble upon this account? Maybe more broadly, we could talk about: How did you become interested in AI influence, or avatars in general? But also, thinking specifically, how did you stumble upon this one and decide this was an area of reporting inquiry?

Hsu: So a mom friend actually texted me one day, and she said, “So I found this account, and pretty sure it’s not a real person. I’m pretty wigged out. What do you think?” And I’ve been writing about AI for a while now. I see a lot of AI. I know what the usual tells are. But I looked at Melanskia, and I was like, This is incredible. Just purely from a technical standpoint, she’s very impressive.

And so I popped the link to her account into the group chat with the other disinformation reporters at the Times, and it just blew up. My colleagues were like, What the hell is this? How did they manage to get Costco looking so real, like down to the labels of the products that she’s holding up in the aisles that she’s walking through? We were kind of stunned at how sophisticated her account was.

Warzel: You have tracked Melanskia down to Melanskia’s creator, or whatever we’re calling this—the creator of a creator. You know, Russian nesting doll of individuals here. Josemaria Silvestrini, who is this entrepreneur you describe as using these AI avatars to promote brands.

Some of them are linked to, as you report, supplement brands, which is its own industry that is dubiously regulated and kind of a little bit of a Wild West in some ways. But what did you learn about Josemaria?

Hsu: My colleague Ken Bensinger, whom I wrote the story with, managed to track him down. We were actually pretty surprised that he was willing to talk. Because for the fact that there are countless posts on social media from people being like, “DM me for tips on how to make tens of thousands of dollars, hundreds of thousands of dollars, on AI influencers,” it’s still pretty murky. And people are not especially willing to chat. But Josemaria was like, “Yeah, dude, let’s talk.”

So I should back up and say that Melanskia is not Josemaria’s only game. He’s got a network of creators that he’s kind of supervising. But he’s not the one who’s creating the avatars. He outsources their creation to other people, and he basically pays them to talk up his products.

Warzel: So a way, maybe, to think about it is: He is both the creator of this supplement product and also, in a way, almost like an agent for these fake influencers for his product. Like, he kind of has this stable of people, which is in its own right very weird. And yes, it was really striking to me that Josemaria—there’s a great picture in the article of him sitting at what looks like a Parisian-style outdoor cafe. He’s smiley, he’s got the laptop. What was his attitude toward this? This is just the brave new world of business. Get on board or be left out.

Hsu: Yeah, basically. I mean, in his mind, he’s building the brand. You know, he’s like, Score. I got a mention in The New York Times; got some earned media out of this. I think for a lot of folks who are creating these AI avatars as, like, influencers or as advertising vehicles, they don’t think it’s sketchy. Even though a lot of them, including Melanskia, don’t disclose that they’re AI. To them, it’s just a more cost-effective, efficient way to market something. If you are able to create an avatar from scratch using AI, you can customize them to do whatever you want, to look however you want, to pitch whatever product you want. You don’t have to pay the fees that you normally do.

While researching this, I went online—always a dangerous thing—but I went online, and I looked at some of the people who are selling tutorials on how to create AI influencers. And, you know, they have taglines like Full guides to making $30,000 a month with AI influencers. There’s one company that’s pitching how to create beauty influencers specifically, which is really weird. Again, because if you’re AI generated, you don’t have skin to use skin care. But the ad for these tutorials says, you know, You don’t have to pay for influencers. Don’t have to pay for studio shoots. You don’t have to pay for the products to be sent out to you. We can make avatars that show a morning routine that can replicate bathroom lighting, that can do simple product close-ups, casual camera angles. It’s really easy to replicate that authentic feeling that makes human influencers so popular.

Warzel: Well, not only that, too, right? I found this great PowerPoint presentation made by this marketer online at the end of the last year that kind of goes through the digital trends of the year. And it’s, like, obsessive. It’s over 300 pages. Right. And it’s talking about what brands, what influencers need to do. One of the things they were showing is, I think they used Wimbledon from 2025 as an example. And it was like, Wimbledon had posted thousands upon thousands of social clips during the tournament—which is only a couple of weeks long, right? It’s just more content than you could ever imagine.

And the idea was like: This is the strategy. It is total bombardment. It’s like many, many times a day. And if you’re an individual influencer and not a brand, you should be posting the same message, like, four different ways, from four different types of locations. Try it at the bus stop, try it in your bathroom, try it at the grocery store. See what hits, right? And it’s just this constant sense of iteration to see what the algorithm is looking for that day, what it’s going to reward—then going back and doing that. And it seems to me that these AI avatars are like … that’s a godsend for this, right? Because it’s simple. You can have all those at-bats just by creating these prompts.

Hsu: Yeah, you copy-pasta yourself, or you copy-pasta your own AI avatar, I guess. But yeah; it’s just so much easier. I’ve talked to a lot of influencers in the past who said the job looks easy, right? You’re just pointing a camera at yourself, and you’re talking to it. But it’s actually really labor intensive, because of exactly what you said. You have to create such a ridiculous volume of content, and it all has to feel fresh. Yeah; I can totally see why someone would be tempted to just make it on a computer.

Warzel: Some of those influencers you’ve spoken with—what do they make of this trend? Is this, like, existential for them? Is this kind of like, Well, you know, they can’t do what we do. Where are their minds around that?

Hsu: I mean, I don’t know that I can exaggerate the number of times I’ve heard the word panic from influencers. Because they’re like, “Oh my God, these computer bots are coming for us.” And it’s true. You know, I think they realize that what they do is so easily done now by a computer that they have to differentiate themselves somehow. I know some of them are hoping that legislation is going to help them. I mean, let’s be frank—it’s not. Legislation, when it comes to AI, is never detailed enough. And it’s always behind, just because AI itself moves so quickly. But there are laws; there’s one in New York that goes into effect in June that requires disclosure. But then you get into the whole thorny issue of whether or not the audience cares if an influencer is AI generated. Let’s say that Melanskia did disclose that she was AI generated. You could really make the argument that A), people aren’t going to see it. You know, we see this a lot with AI, where even if there’s some sort of note somewhere that says, “This is an AI-generated piece of content,” you go into the comments and people are like: “Whoa, this is crazy. I can’t believe this happened.” Or, “Wow, you’re so gorgeous. What’s your number?” Or the second possibility—and to me this is a really scary one—is that people just don’t care. They’re like, “We know you’re a collection of pixels, but we like what you’re saying. We like what you’re showing us, and we are influenced.” And I’m seeing that a lot.

Warzel: So I saw a video of this blonde woman who looked like a soldier named Jessica Foster, who at scroll speed fooled me, right? I didn’t really pay much attention to it. But then I went back up, and I was like, Wait, Donald Trump’s meeting with, you know, this soldier in the Oval Office. It kind of, you know—Spidey Sense went off. And the post I saw that I felt really captured this well was, “Well; what’s the difference between this fake soldier woman and some of the other people who are real who some of these same men will see on Instagram? Both are unattainable in the same way.”

As in unattainable, not like in a dating sense, but in a, “I’m never going to meet this person,” right? Like, I’m never going to live the life. I’m never going to have access to the spheres of where they are—whether that’s on a battlefield or in the Oval Office or in Hollywood or in a penthouse by a pool. And I think that that’s a very real thing for a lot of people, right? Like the influencer is meant to be an avatar—even if they’re real—for a life that is aspirational. An AI avatar can do a lot of that, and it’s going to be just as hard, you know, for someone to feel like it’s attainable.

Hsu: Yeah. I mean, influencing, at its root, is an exercise in wishful thinking, right? So it doesn’t matter if, you know, you’re hoping fruitlessly that you’re going to be this real person or this AI-generated person. My team normally covers disinformation. And we came across this idea a lot when we were covering the hurricanes a while back in North Carolina, where there were a lot of AI images being posted of the devastation. There was one showing, I think it was a little boy, I want to say, who was sitting on a raft with a dog. Drenched, crying.

So this image gets posted by a local Republican official. And immediately people on X are like, “This is AI generated; this is not real.” And she responds in kind of a stunning way. Which is to say, “I don’t really care where this image came from.” Like, “It hurts my heart,” or “The feeling is real.” And what’s happening a lot on social media, just because there’s so much content, is that people are getting fatigued. They are so exhausted at having to parse what’s real and what’s fake that a lot of them are just saying, like, “If it makes me feel a certain way, that’s all I care about.”

Warzel: But do you think that fatigue goes the other way? With people saying, You know what, I’m becoming generally over the influencer thing. Because of the fact that it’s like, if you’re not even going to invest the bare minimum humanity into this product—that is, this product is used by actual people in some capacity—I’m going to bow out of it. Do you think it could sort of erode and taint the whole idea of the influencer-like ideal? Or do you think maybe it just supercharges it?

Hsu: No; I think you’re totally right. Aerie, this bra company, posted in October this pledge that they were only going to use real people. So they said, “A few years ago, we stopped retouching our models. And now we’re going to pledge to never use an AI body.” And that post was by far their most popular post of the entire year. I think you have a lot of brands that are catching on that, you know, people are over AI influence.

But then, on the other hand, you have avatars like Aitana Lopez. She is described as an influencer who’s based in Barcelona, who’s a fitness girl. She has pink hair. I mean, she discloses herself as being AI generated, but she has nearly 400,000 followers. She has been posting photos of herself in, like, Schiaparelli dresses at Paris Fashion Week. She’s got a brand deal with Alo Yoga. You know, her creators have been quoted saying that she makes up to 10,000 euros a month. So I think there is some resistance to this trend, but they’re still really popular.

Warzel: Do you feel like some of this is just a literacy issue? Or do you feel like, again, it’s really a change in what we expect, what we consume, and a lack of really caring?

Hsu: So I think that’s a really good question. I think the answer is—it’s a little bit of both. I did a story late last year where I shadowed a high-school class, and they were learning about disinformation and media literacy. But they had really interesting perspectives on AI influencers, which they independently brought up to me. Where they said, “You know, some of them are really hard to identify as being AI.” You know, a couple of them mentioned after Sora came out, there were a series of posts that featured Jake Paul, the YouTuber, supposedly embracing his queer identity and like doing makeup videos, whatever. A couple of the kids were saying, “We were almost duped by that.” And so for them, in part, it’s a media-literacy issue. To know that AI is capable of producing stuff like that is something that they want a lot of their peers to be on top of. But on the flip side, they raise a perspective that I think journalists often forget now, just because AI is so crazy—which is that AI is really impressive. Like, it’s used to create incredible things. So a lot of these kids are themselves experimenting with Google Veo or Nano Banana Pro or Seedance. And, you know, they’re playing. So to them, having the possibility of influence via AI avatar is like a fun thing.

You know, one of them mentioned this influencer studio. It’s basically like a modeling agency, but for AI avatars. It’s from a company called Higgsfield. They have a program called AI Influencer Studio. And you can pick from so many different options. It’s like Sims for the new age, right? You can choose whether you want your avatar to be human, an ant, or an iguana or an elf, right? You can choose genders, like trans man or nonbinary. Can choose skin color, eye color. You can choose, like, a skin condition that includes like burns and dry, cracked skin. They have settings like forked tongue, big or small horns, or fish skin, right? So I think, to a certain demographic, like messing around with this is a really great time. It’s just, they’re doing it for kicks.

Warzel: Right. And that makes sense. I think it is important to keep in the context here. There’s sort of two buckets of this, right? One is that bucket of being able to play around. That these are tools; these are forms of AI. You know, detractors and critics will probably bristle at the phrase, but there is an art form to this. A creative expression formed to all of this. And I think that we can’t discount how it can be fun, and it can just be a different way for people to move through the world with these avatars, right? And then I think even in the brand space, in the selling space, right? We’ve got Tony the Tiger; we’ve got people who sell things that aren’t real, right? And yes; it’s obvious that there’s not a talking tiger who’s real and selling you cereal, or whatnot. But there is this idea of, like, “characters sell things.” They always have; they always will. Brands have mascots. And so I think there’s precedent, probably, for a lot of this stuff. And there’s a lot of this that isn’t pearl-clutching panic.

And yet I think what interested me about your story that you did with Ken, and the reporting, is the way too in which it overlaps with this world of less-than-regulated supplements, and things like that, right? And so I think, that seems to me to be the concern here, right? That in this moment—where we’re all figuring out how to understand this, how to read this, up our literacy on all this—you have this group of people who are moving in. And they’re actually trying to take advantage of the fact that there is a lot of confusion in this space.

Hsu: Oh yeah. I talked to Tim Caulfield, who is a professor of health science up in Canada. He put it pretty succinctly, where he was like: People have realized that AI avatars are a great and easy way to make money. And now the scammers are like, “Hey, let’s hop in on that.” And the wellness space has always been a magnet for scammers.

I’m a mom, and I remember both times I was pregnant, the absolute amount of insane content I would get served on social media. And the idea that fake people who seem real could be selling that sort of stuff to me, with like a voice of authority, is pretty scary. Because I think—more so than fashion or beauty—wellness stuff, people are really willing to trust. Because most people don’t have the kind of science background you would need to really parse through what is snake oil and what is like a legit thing.

And so, if they see someone who seems like they know what they’re talking about, who seems to be talking from experience, they’re more inclined to trust. So wellness scamming has a long, illustrious history of bullying a whole lot of people. And you’re right. The presence of AI avatars in that space is only going to make that worse.

Warzel: So when you were doing the reporting, and it led you all to this guy, Josemaria—who’s sort of the keeper of this, the guy running the brand, but not making the avatars themselves—did you all get into how these avatars are created? There’s obviously these programs that exist. But the people who are sort of mercenaries for some of these folks, right? Did you get a sense of where the hotbeds of this AI-avatar creation are?

Hsu: Because it’s so easy to make an AI avatar, you don’t even need to be working in conjunction with other people. You know, this can be a do-it-yourself thing. There are guides all over the internet teaching random people how to do this. I could probably do it if I felt like it, pretty quickly. Because the regulations are so confused, and they’re so lax, not a lot of the platforms that offer AI generation are going to stop you from creating an AI avatar. ’Cause there’s nothing inherently illegal about creating a fake person who says certain things about a certain product. You could run afoul of rules about scamming, but the platform isn’t going to be able to monitor that. It’s not in their interest to put in a lot of resources into overseeing how people are using the characters that they’ve created with the program.

Warzel: Well, and you mentioned, with regulation, that in December, the governor of New York signed the nation’s first legislation, and that explicitly required this disclosure of quote unquote “synthetic performers” in certain advertisements. You said it’s kind of DOA in some sense, or it’s just not really going to have this impact. Can you say more about that?

Hsu: Look, I think it is helpful to have the legislation, just because it sets a precedent for other attempts to regulate. I think the more practiced lawmakers get trying to regulate this space, the better they’re probably going to get at it. It also demonstrates that the authorities care, that they recognize that there’s a problem. Even if the way they’re going about issuing consequences is a little bit hazy. I mean, it’s just—so many of these creators are anonymous. A lot of them are operating from outside the country. I mean, this has been a problem with all sorts of AI legislation, whether it has to do with deepfake porn or political deepfakes. Dealing with commercial scammers is not super high on the average legislators’ priority list. The deepfake porn is the area that most folks are really, really up in arms about at the moment.

Warzel: But even there, it’s also a volume game, right? It’s like, if you were going back to the old paradigm of people scamming, these things are conducted at human scale. And now at artificial-intelligence scale, and with agentic AI and the swarms of AI agents and doing stuff. It just sort of exponentializes the ability to do this stuff. So it seems like it goes from whack-a-mole to, you know, whatever … some inhuman game of whack-a-mole.

Hsu: I mean, I’ve talked to a lot of victims of deepfake porn or AI-generated threats. And some of them have said that they’ve complained, and they’ve managed to get the platforms to either block or remove the responsible accounts. And then another account pops up a day later and starts targeting them again. Like it’s turbo whack-a-mole; exactly what you said.

Warzel: Where do you think this goes from here? Because I get this sense that we’re in … I can talk myself into it two different ways, right? That this is the first inning of a very long game, the likes of which are going to get stranger and stranger and more dystopian. Although maybe we grapple with it. And the other, I think, too, is this idea that yes, the world is getting weirder, more unpredictable. There’s more tools for scammers and folks like this. But also this—like we were talking about earlier—this exhaustion. In, you know, being bombarded by this digital stuff. The internet becoming less and less human, and people dropping out, or not finding themselves all that interested in it.

When you think about what you’re trying to anticipate and what’s coming next, do you fall into one of those two camps? Do you have a different thought about where all this goes?

Hsu: Given our track record as humans, I’m not incredibly optimistic about us, you know, putting our foot down collectively as a society and saying, “Stop this AI nonsense.” I do think we’re going to run into a lot of issues with societal trust. I mean, that’s already happening. My colleagues and I just wrote a story about the “liar’s dividend”—which is the phenomenon that happens when the prevalence of AI makes it so that people can more easily discount actual footage, real footage. Which is what’s happening with the proof-of-life video that the prime minister of Israel has had to circulate.

Warzel: Explain that for a second, for people who will be less informed about this idea that Israeli Prime Minister Benjamin Netanyahu is dead. Explain that just for a second.

Hsu: Oh God; how many hours do we have? So in a nutshell, Netanyahu gives a speech that is recorded. And in the reshared versions of the recording, in certain frames he kind of looks like he has six fingers on one hand. Now, extra fingers is, or was, a pretty common tell that AI was involved. Now, AI’s gotten much, much better, so that’s not really the case anymore. Versions of that video start circulating, where people are like, My God, he has six fingers. This is AI generated. Obviously, he’s dead. Because we always jump to that conclusion now—that the world leader is dead. So amazingly, a few days later Netanyahu posts on his social platforms a video of himself at a cafe outside of Jerusalem, gesticulating very clearly with his five-fingered hands. I mean, this is, it’s a proof-of-life video. And, as far as I know, it’s the first proof-of-life video to directly address being deepfaked from a world leader, especially one as prominent as him.

Okay. So he posts this video. It is verified in several different ways. The café itself posts its own set of images, separately from Netanyahu’s, showing him ordering and talking to people. You know, several deepfake professionals analyze that video frame by frame. They’re like, “We don’t see any signs of digital manipulation here.” Regardless, the internet goes, “This proof-of-life video is also AI generated.” And you have people, some of them with millions of followers, post copycat videos of, like, the new leader of Iran doing the same things that Netanyahu did in that café. Or they show Netanyahu wearing a sports jersey in the café, just to prove how easy it is. And so the narrative just spiraled. It’s like people don’t trust the proof that was provided in response to the initial distrust of video proof.

Warzel: Doing reporting on all this disinformation stuff in like 2016, 2017, 2018, there was so much of this idea of like, “You think these Photoshopped images are bad? Just you wait till the AI stuff comes.” And like, the AI stuff was in the realm of the really bad videos of Will Smith eating pasta where his mouth detaches from his body. And there was sort of this, like, “Okay, yeah; I can see that happening.” But this idea of like, “No, the lines of reality will blur so fully that it just will be a free-for-all.” And I think that there’s so many people that received the articles that I wrote about this, or the reporting I did, and other people out there were like, “Okay, that’s really alarmist.” And I think it’s fair to say that we are just actually living in that future. Like, that was a, like, 100 percent success-rate comparison. And I think that it’s not where I think a lot of people would go with, you know, an immediate jump from AI avatars to this. But you’re totally right. The more that this technology comes into our lives in mundane ways, the more we expect to see it in these unprecedented, really high-stakes ways. And the more that people can basically say whatever they want to say and have plausible deniability.

Hsu: Yeah. People are so desensitized to it, too. We keep writing about how bold a lot of social-media creators are becoming now. Because, I mean, the [Trump] administration is a meme factory. Like, we have a president who communicates through AI images and digitally warped images of real events.

Warzel: I believe they refer to them as bangers, so that’s the White House term of art.

Hsu: Right. My God, I got a comment back from the White House when I was writing about this photo of a protester in Minneapolis who had been taken into custody. You know, the original photo is posted by some branches of the government, showing her, you know, pretty composed. She’s walking with an agent. And then the White House posts a photo of her with her skin darkened, and she’s sobbing. And I reached out to the White House for comment on this, and they’re like: “Justice will continue to be served. The memes will continue to be served.” Or something to that effect. And I was like, So this is our communication method now.

Warzel: Yeah; the paradigm is completely shifted in that sense. This is a little bit of a swerve here before we land the plane. But something I was thinking about is, you said: “I think I could create one of these avatars using these programs,” right? And I think I believe you. But could I get it to be, like, highly influential? Like if I had this thing to play with, is there still a skill game at this? It isn’t the slam dunk that people think. It’s actually like, these people are just really good at the game, and they just happen to do it with a costume on.

Hsu: Short answer is yes. I have a long answer. I’m going to tell you about my journalism white whale, which is that I have tried for years now, I think, to get a mention of one of my favorite movies into a story about AI. Which is the masterpiece called S1m0ne. Do you know this movie?

Warzel: I don’t think I saw it.

Hsu: I mean, honestly, it was not popular. I don’t think many people saw it. But for some reason, I love this movie. Mostly because it keeps echoing across my work. So in short, Al Pacino plays a director who’s down on his luck, because he’s not being successful. So he essentially creates an avatar named Simone, who he convinces everyone is a real person. Simone goes on to win Oscars. She, like, runs for office and wins. And Al Pacino’s character at some point is like, My God, I’ve created a monster. He puts all of the CDs that Simone is imprinted on, and he tries to like bury them in the ocean. And he gets, like, accused of murder. It’s a really convoluted, messy story. But the fact that an AI or a synthetic character manages to convince the entire world that she’s real, and she is able to exert huge amounts of influence, has always stuck with me. And I think increasingly more so, because it seems like something like that could happen now. That you could get someone who understands the way media works, who understands the way Hollywood or social media or audiences in general work. And you could easily have someone who creates a character that really is compelling to a lot of people.

Warzel: So real quick: What I want to do, I want to talk about a specific Melanskia video. The aforementioned rotisserie-chicken video with the caption “Most people buy this every day.” And I would love if you could walk me through the tells here. How people who are just scrolling through their feeds and stuff are going to be able to, like, distinguish this. You know, I’m seeing five fingers, for example, so it’s not that. What are some of these tells here?

Hsu: This is a little gross, but if you look at the way the chicken is dripping … I don’t think rotisserie chickens drip quite so lusciously. So there’s that.

Warzel: It is disturbing.

Hsu: Yeah, right. I don’t want to look at that more than I need to. One of the biggest tells for Melanskia specifically is that if you go to the grid of her overall account, you notice that she’s always kind of positioned exactly the same way. She’s always looking at the camera exactly the same way. She makes very similar facial movements and hand gestures. That tends to be a tell, but it’s hard to notice that in a single video. Other—sorry, go ahead.

Warzel: She’s always kind of lit with the same golden-hour light; like that’s what I see from the grid. It always looks like it’s 5 p.m. in the summer, you know. And she’s outside, and sometimes it looks like she’s … I mean, she’s inside in some of these. She’s like in Costco.

Hsu: I mean, that’s actually a really good tell, is lighting. Often with AI influencers, the lighting isn’t natural. Like, it kind of looks like they’re being lit from all sides instead of just from one direction. If you look at some of the older videos, she looks a little bit different.

Warzel: Oh wow; yeah.

Hsu: That tends to happen with a lot of the longer-running accounts. I don’t know if this is the case specifically with Melanskia, because she doesn’t really do super close-ups. But sometimes you’re able to look into the irises, and the reflections are different in both eyes. If you look along the hairline, often it’ll look a little blurry, or a little out of whack. Um, if there’s audio—audio is really, really good now, audio deep-faking, but sometimes they don’t breathe like a real person does. There aren’t as many … like, for example, the way I’m talking. I use a lot of ums, a lot of filler words. AI avatars do that less. But, of course, all of these things you can deal with if you’re a really good prompter.

Warzel: Yeah; I’m just watching this chicken drip. No liquid behaves that way. Anyway, sorry about that, to derail it. But no; those are all great. In some ways, I’m half heartened by all of this, right? Because I can’t tell you, always, what it is. As you’re walking through that laundry list, I’m like, Yep, I see that. The hairline; that one was novel to me. The zooming in, looking at the irises. But there is just something that my brain still recognizes as suspect, right? And I’m also worried, because I’m like, Is this the last glimmer? Like, My brain—is this the last flickering of this instinct before I lose it?, you know.

Pre this conversation—like, cards on the table—it’s sometimes hard for me to get really fascinated by AI influencers. Because I’m just like, Yeah, that’s not for me. That’s just not a thing; I’m not interested in engaging with an influencer who’s not human. And so how could everyone be? And I’m like, it seems really plausible to me in the same way. It’s obviously very different. But if you were to tell, like, someone in 1989, “This guy Donald Trump’s gonna be president, right? And people are just gonna be like, ‘He’s a genius and a strategy master,’ and all this stuff.” Right? People would be like … Okay.

That’s what I’m thinking about as you’re saying this. Yeah; it’s not gonna happen tomorrow, or anything like that. Like, we’re not gonna have, you know, whatever, President Simone. But I do think it’s really interesting to think about all of that, and that kind of dynamic, in a world where this becomes more normalized. And also in a world where maybe that North Carolina disaster-politician ethos of “I don’t care that it’s not real; it speaks to me.” If those things marry in a way—culturally, politically, whatever—I do think it really brings up this question of like: Man, are we gonna get to the place where there’s gonna be influential people? Who even have gained people’s trust in ways that right now seem really absurd? It seems really plausible in that sense.

Hsu: Yeah. You know, I completely agree. So in my reporting, I think a lot about, like, identity and anonymity. Because so much of the really sketchy shit that happens, in my line of work, is done anonymously. And, you know, recently, Banksy, the artist, has appeared to have been identified, right? As, I think, some 50-something-year-old man in the U.K. It just got me thinking that, you know, for years, this guy—whom no one could really identify—was out there, like, changing the art world and commanding ludicrous prices at auction. And now that he’s been identified, is that going to change any of that? Do people really care? Or is it just the content that they’re interested in? And I don’t know if this is a torture link back to AI avatars, but I think the same question is valid. Are we at a state where we really don’t care who’s behind a major cultural figure? And it’s really just the image that they’re putting out, or the product that they’re putting out, that is more compelling to the audience?

Warzel: Tiffany, this is, I think, a good place for us to leave it. I’m impressed with where we ended up getting to here from this. And I really feel like now I’m to go and actually contemplate an AI-avatar politician and stare into the abyss. So thank you for that.

Hsu: Hey, let’s leave journalism and make some real money. Seems easy.

Warzel: I think it’s time. I think I found my off-ramp.

Hsu: Let’s do it. All right. Let’s make it happen. Thank you.

Warzel: This is great. I appreciate it.

[Music]

Warzel: That’s it for us here. Thank you again to my guest, Tiffany Hsu. If you liked what you saw here, new episodes of Galaxy Brain drop every Friday. You can subscribe to The Atlantic’s YouTube channel, or on Apple or on Spotify or wherever it is that you get your podcasts. And if you’d like to support this work and the work of the rest of my colleagues, you can do so by subscribing to the publication at TheAtlantic.com/Listener. That’s TheAtlantic.com/Listener. Thanks so much, and I’ll see you on the internet.

This episode of Galaxy Brain was produced by Renee Klahr and engineered by Miguel Carrascal. Our theme is by Rob Smierciak. Claudine Ebeid is the executive producer of Atlantic audio, and Andrea Valdez is our managing editor.

An Incredibly Weird Time to Be Alive

2026-04-07T19:56:00-04:00

Seeing the Earth from space will change you so profoundly that there’s a term for it: the overview effect. The extreme minority who have had the privilege describe it similarly. You see something that you were never meant to see, namely the Earth just sitting there, with the entire universe surrounding it. Gazing upon the blue marble, surrounded by its oh-so-thin green layer of atmosphere, the auroras flickering on the fringes, is not merely awe-inspiring but something of a factory reset for one’s sense of self. Almost everyone tears up at the sight.

“You don’t see borders, you don’t see religious lines, you don’t see political boundaries. All you see is Earth, and you see that we are way more alike than we are different,” Christina Koch, one of the four astronauts on the Artemis II mission, told NASA recently. Jim Lovell, describing the view on Apollo 8 from the dark side of the moon back in the late 1960s, told Chicago magazine that he could put his thumb up to the window, and in that moment, “everything I ever knew was behind it. Billions of people. Oceans. Mountains. Deserts. And I began to wonder, where do I fit into what I see?”

Where some see immeasurable beauty, others see fragility. Marina Koren previously reported in this magazine that, upon seeing the Earth from space, one astronaut “became absolutely convinced we would kill ourselves off between 500 and 1,000 years from now.” Famously, the actor William Shatner has written that his brief experience looking at the Earth produced a profound sadness. “What I was feeling was grief, and the grief was for the Earth,” he told Koren in 2022.

I’ve never been to space, but for the past few days, I’ve oscillated between these emotions—awe and despair—as NASA has continued to post photos of the Earth and moon from Artemis II. Yesterday, the Integrity spacecraft came within 4,067 miles of the moon during its lunar flyby. For 40 minutes, it lost all contact with humanity. At one point they were 252,756 miles away from Earth—the farthest from the planet anyone has ever traveled. For seven hours, the astronauts—Koch, Reid Wiseman, Victor Glover, and Jeremy Hansen—were able to gaze upon a part of the lunar surface previously unseen by human eyes. According to NASA, the astronauts took roughly 10,000 photos, which feels perfectly proportional for such an occasion.

A few of these photos—some taken before the lunar pass—have messed me up pretty good. A photo of the Earth appearing to set behind the moon. A picture, taken through a window of the Orion spacecraft, revealing the tiniest crescent Earth growing smaller as the capsule heads toward the moon. As one caption on the photo notes, “The Earth is illuminated by the blackness of space.” I’ve experienced these photos the way I experience most media: through the puny screen of my phone, with the awesome, life-affirming images sandwiched between updates about a golf tournament, oil prices, the MLB’s new automated ball-strike system, and reports of the U.S. president threatening the civilizational destruction of Iran.

On a good, calm day it is hard to know what to make of photos that show, in no uncertain terms, that every single thing you will ever and could ever know is simultaneously galactically insignificant and unspeakably beautiful and precious. Today, the world held its breath waiting for the 8 p.m. eastern deadline Trump set for Iran to agree to a deal to reopen the Strait of Hormuz. If his terms weren’t met, he posted this morning, “a whole civilization will die tonight, never to be brought back again.”

Trump’s threats triggered denouncements from Democratic lawmakers as well as the podcasters Tucker Carlson and Alex Jones, and incited no small amount of panic from people who have interpreted Trump’s post as a suggestion of nuclear warfare. Then, this evening, an hour before the deadline, Trump announced a two-week cease-fire deal, which Pakistan helped broker.

Trump’s bluster, no matter how serious, has always been impossible to parse. (He’s famous for chickening out, backpedaling, or pretending like he never said what he said.) Yet one way to view our current age is as a series of existential reminders, be they nuclear proliferation, climate change, or pandemics. In Silicon Valley over the past half decade, civilizational extinction at the hands of hypothetical technological advances has moved from the realm of pure science fiction to a marketing tactic to an immediate concern for a subset of true believers. Humans may not want to die, but as a species we seem eager to invent and tout new ways to threaten our existence.

And yet at the very same moment, four flesh-and-blood human beings are hundreds of thousands of miles away taking pictures of our delicate little world. Their mission and their photos remind us of something else entirely—of a yearning to learn, to explore, and to band together to become something greater than the sum of our parts. If Trump’s claims of mass destruction represent humanity at its smallest, weakest, and most cowardly, then those who are gazing upon our planet right now from afar represent the best of what we have to offer. How else to hear these words from Koch:

We will explore. We will build. We will build ships. We will visit again. We will construct science outposts. We will drive rovers. We will do radio astronomy. We will found companies. We will bolster industry. We will inspire. But ultimately, we will always choose Earth. We will always choose each other.

As Lovell looked down at the Earth in 1968, an old saying popped into his head: I hope to go to heaven when I die. Then he realized, “I actually went to heaven when I was born.”

There is something disorienting, horrible, and somehow fitting in the timing of all of this. That one man with the means to do it would threaten destruction of a part of our planet at the same moment its beauty and fragility are on full display. We are, in this tense moment, living with our own overview effect. Four are watching from afar. But the rest of us are watching too—left to reckon with our own place on the pale blue dot, reminded of all the ways we might die, and all the reasons for which to live.

*Sources: NASA; Space Frontiers / Getty; Chip Somodevilla / Getty.

Is AI Going to Turn Us All Into Middle Managers?

2026-04-03T13:00:00-04:00

Subscribe here: Apple Podcasts | Spotify | YouTube

How is AI changing the way we work? This week on Galaxy Brain, Charlie Warzel is joined by Johnathan and Melissa Nightingale, two experts in management and leadership training. They discuss how chatbots and AI agents are winding their way through the workforce, offering a firsthand view of how companies are (and aren’t) adopting AI tools. The conversation covers the gap between AI hype and what’s actually happening in offices. It also touches on how overreliance on AI tools may be making bosses worse at their jobs, and how work may be one of the last bastions of sustained social connection in a period of cultural alienation and isolation.

The following is a transcript of the episode:

Johnathan Nightingale: It turns out that humans really care about doing work they believe in, with people they care about. And when you hollow those things out, people have these emotional responses to it that I don’t see predicted by the marketing materials from the AI companies.

[Music]

Charlie Warzel: I’m Charlie Warzel, and this is Galaxy Brain: a show where today we are going to talk about work. Trying to talk about jobs right now—how we work, what that work means, what the future of white-collar work looks like—it is just extremely difficult.

We all seem to be situated in this very confusing moment, one that I think is captured very well by my Atlantic colleague Josh Tyrangiel’s story. It ran with this rather ominous headline: “America Isn’t Ready for What AI Will Do to Jobs.” The piece is great, and you should totally seek it out. But the illustration of the piece I found quite apt.

It’s of this man; he’s in a tree. His eyes are two dollar signs, and he’s smiling while holding a chainsaw and cutting into the branch that’s supporting his own weight. This in many ways is the vibe of 2026. This feeling that this certain subset of people motivated by profit and efficiency are conducting an experiment that, if it succeeds, is not gonna just rewire the economy forever but change the very nature and essence of labor.

For the last few years, since the arrival of chatbots, the AI conversation around work has been some version of this. The tools are useful in automating busywork and drudgery. They’re getting better.

And so what does that mean for jobs? Well, AI executives have been issuing dire warnings. In 2025, for example, Dario Amide—the CEO of the AI company Anthropic—told Axios that AI could drive unemployment up 10 to 20 percent in the next one to five years and wipe out half of all entry-level white-collar jobs.

Meanwhile, those who are running companies seem quite eager to unleash this technology on knowledge work—labor force be damned. They’re driven by profit incentives and a good amount of FOMO. AI is the future. Get on board or be left behind. What’s your AI strategy? In some cases, the fastest way to show results is to simply reduce head count.

Workers, especially young workers, are concerned. According to the Federal Reserve Bank of New York, the unemployment rate for college graduates ages 22 to 27 ballooned to 5.6 percent at the end of last year. And as The New York Times notes: “For those who are employed, more than 40 percent held jobs that do not typically require college degrees, the highest level since 2020.”

You can feel the weirdness in the economy right now. This fear of a kind of job-market stagnation, but no exact sense of what is happening on the ground. How much of all of this is actually AI driven? Simply put, there are huge fears here that AI is not only changing the way we might do our jobs, but it might be changing how we get them, whether we can keep them. AI executives are out here arguing that most of us, no matter the job, are destined to become middle managers for a host of AI agents. And you can take that with a grain of salt as just another tech CEO prognostication. But if there’s any truth to it, it would represent a massive change. And the concerns go well beyond the economy into something much more existential. What does it mean to be a human in a world that all of a sudden doesn’t value human labor in the same way?

So are we all destined to become middle managers? Is AI really ripping through the workforce? What is the value of work in this strange economy? And how can people survive—maintain dignity, human connection, all of that—in a world where decision makers driven by dollar signs are pruning the trees for every possible efficiency? Even if we’re all just sitting on the branches?

Johnathan and Melissa Nightingale are here to help me answer some of these thorny questions. They’re the founders of Raw Signal, a leadership and management training firm. Johnathan and Melissa have worked with thousands of executives and managers and companies across tech and other industries, and they’re keen observers of the ways that modern white-collar work and workplace communication is broken. But they also offer this clear vision of how work can stay human and humane. They join me now to talk it all through.

[Music]

Warzel: Melissa and Johnathan, welcome to Galaxy Brain.

Johnathan Nightingale: Thank you.

Melissa Nightingale: Thanks, Charlie.

Johnathan Nightingale: It’s lovely to be here.

Warzel: So it is a weird time in the economy, especially in America. We’re in this low-hire, low-fire labor market. There’s this very amorphous and pressing fear right now about artificial intelligence taking or threatening jobs, especially entry-level jobs. And the labor market just looks increasingly grim and feels increasingly grim to younger workers. The vibes—they’re off, they’re not good.

I wanted to start just by asking you all: You’re talking to businesses, to people. You’re on the ground. What are people telling you about the vibes right now? Report on the vibes for me.

Melissa Nightingale: I love “Report on the vibes.”

Johnathan Nightingale: “The vibes” is such a great place to start. Because I don’t know if you remember, but it wasn’t so many years ago that companies were appointing “chief vibes officers.”

Melissa Nightingale: Do you remember this? Like, in 2022?

Warzel: I don’t really, no.

Melissa Nightingale: Okay; there’s a very weird future-of-work moment where everybody was, like, a future-of-work thought leader in 2022. And The New York Times reported they had a big story about chief vibes officers being like…

Johnathan Nightingale: The new jobs of the new economy.

Warzel: Man, I missed out on that. I would love to be chief vibes officer.

Melissa Nightingale: But that was the new hot title.

Johnathan Nightingale: Because, like, in 2022, like everybody was fighting. I think junior engineers were getting, you know, 100, 200, 300,000-dollar offer packages because everybody was starved for this tech talent. And that was the story of the moment—“Wow, how much worker empowerment there is.”

Melissa Nightingale: And so, we’re not trying to make the vibe sad. But it is worth starting with: Where were the vibes as we head into … like, where are the vibes now? So 2022: still relatively hot labor market; still a lot of competition for talent. Particularly junior talent; particularly junior engineering talent.

Johnathan Nightingale: And you could tell that senior leaders were pissed off about them. It was too expensive. These people were too entitled, right? Like, “chief vibes officer” makes good buzz. But you could tell in early 2022 that this was about to get some pushback.

Melissa Nightingale: Remember, this was the earliest version of return-to-office mandates. People saying: “We went home. We did amazing work at home. Why do we have to go back? And also, like, if you make me go back, I have another $300,000 job offer lined up tomorrow.”

Johnathan Nightingale: Just walk across the street. And so, it’s funny because people talk about AI and all the layoffs. And there’s been, you know, half a million layoffs in the last several years, and technology workers and stuff like that. Those layoffs started before the first ChatGPT release.

Melissa Nightingale: What it did was reset the market, right? It recalibrated. Like, you had folks who had that sort of early engineering role. And they got a little bit more skittish about, Should I leave to go across the street, or should I stay put? Should I feel grateful that I have a job, that I have this opportunity? And you started to see people feeling a little bit more nervous, and a little bit more uncomfortable around the market overall.

Johnathan Nightingale: These days, AI has sucked all the oxygen out of the room. And people are like: “That’s all driven by AI.” I’m like: no. First of all, November ChatGPT couldn’t count how many Rs there were in strawberry. That’s not a reason to turn your whole business upside down. But also, the layoffs happened six months before that too. It’s really been part of a pattern of executives in some organizations reasserting power and making sure that, especially, junior workers lose that sense of entitlement. I think vibes-wise, that’s happened. They succeeded in that.

Warzel: It was such an interesting moment. I wrote back in—I think it was 2021, but it might’ve been 2022—in that time period of a hot labor market. Also of, as you say, some real worker empowerment. I remember writing this piece that was like: “Do workers, do people even want a career?” Right? Like, do the young, Gen Z people coming up, do they even want a career? They’re questioning the idea of the standard thing, because there’s just so many different options. And maybe I don’t want to work the way that my parents worked. And to compare that feeling to now, where it’s like “Could I even get in the door to have a career, or am I going to have to figure out something else completely different?” I think that that’s extremely stark in terms of that shift. What are people telling you on the ground now in terms of this moment? Obviously there is that sense of precarity with workers. But in terms of that force, of generative AI, how are people feeling about it?

Melissa Nightingale: Johnathan and I both come from tech, right? We’ve both been working in tech since the first dot-com boom, all the way on through. And we work with a lot of organizations with tech leaders. And what we’re hearing from folks is—on the one hand, you sign up for tech as your industry and as your career, you like working on the cool new stuff, right? Like, we are an industry that loves our toys. We love innovation. We love sort of taking things out and experimenting. Sometimes they last; sometimes they don’t last. But as an industry, if that doesn’t get you lit up, you’ve picked the wrong job.

But what we’re hearing from a lot of folks is that the day-to-day of “We’re playing with these tools” is no longer lining up to “What are we supposed to be doing here?” And that playing with the tools has become sort of an end into itself. And a lot of folks are finding: “We’re using them, but I don’t know—what are we running toward?”

Warzel: Do you get the sense that this is like, if we’re going down the chain: This is executives high up on top, see a thing. They’re reading a lot about it; there’s a lot of hype. They feel, Okay, more than anything, I have to make sure that we do not get left behind here. And then the sort of middle-manager layer is feeling that it’s forced on them. Or is there sort of, broadly speaking, a lot of enthusiasm—but there’s just not enough time to figure it out?

Melissa Nightingale: All of the above. Like, we met a lady who’s a really skilled executive, and she was working in an organization where she’s like: “I came in as a fixer, right? My whole job was fixer. And I was really excited about it. Like, the opportunity was cool. Came in, doing the work. And starting to see the impact of that work, right?” And she’s like, “And then my CEO got really excited about GPT, and I started sending things that were strategic plans for my division, for my department. And what started coming back from my CEO—who I report directly into—wasn’t from him. It was clear he hadn’t read any of the plans that I was putting forward. He just pushed them through and said, Generate an email in response to this.” And she’s like, “I went from being so excited about the turnaround potential for this business, for this organization, and for my department and team, to feeling really sad. Like fundamentally—just having a very hard time figuring out, What am I doing if I’m putting in all this effort?”

Johnathan Nightingale: When you look at a story like that—that’s a management failure. It can take one of a couple flavors. One thing you can say is, “That person’s work is now useless because GPT replaces it well enough. And so that person should have just been given a firm handshake and a goodbye package and sent out the door.” Or you can say “GPT can’t replace human ingenuity at the senior levels.” And so there’s a real dereliction there, because you had a motivated, engaged senior employee, and you burnt them. In either case, that’s a management failure. A thing that’s coming up a fair bit is that people look at what code generation can do in LLMs today. And they sort of do the “well naturally”: Well, naturally, in time, it will write good poems. Well, naturally, in time, it will be your accountant. And well, naturally, in time, it will manage as well or better than most people do. It’s just this leap, this linear leap, that is not borne out by what we’re seeing on the ground today.

Warzel: Do you feel like that’s a broader trend of, especially on the executive layer—and I’m not trying to paint people as caricatures here—but this idea that, especially with these higher, more like “vision strategy” jobs, where there is a lot of busywork involved in that sense of communication. That, you know, what I’ve heard is that AI is a perfect CEO, right? Like, it could be—it’s just sort of broad, broad pronouncements—being able to speak with maybe more confidence than is earned or deserved, right?

Johnathan Nightingale: Incredible executive presence.

Warzel: Great presence in that sense. And so, naturally, people higher up on the end of the management chain might be enamored with it or the ability of it. Do you feel like CEOs are overly AI-pilled right now?

Johnathan Nightingale: I think they’re a vulnerable group. I think one of the challenges with being a CEO is that even an incredibly effective CEO shouldn’t know more about every function in their business than the people working those functions do. Right? If you’re an expert engineer and you become a CEO, you might know a lot about engineering.

Melissa Nightingale: But over time, you’ll actually know less than the people who are typing on keyboards all day.

Johnathan Nightingale: And your head of sales should know more about sales than you do. Your head of marketing should know more about marketing than you do. Right? Like, that part of the job of building a senior team is to make sure those people are, one, lighting you on fire with great ideas, and two, are credible leaders for their own functions in a way that you couldn’t be. Because a person can’t know everything about everything. And so, it’s tempting from that seat to flatten a lot of that work, and to say “If an AI can do that work 80 percent as well, maybe I have to spend some more compute over there in order to get the result I want. But, like, do I even need a marketing department? Do I even need an engineering department? Do I even need a finance department?” They’re vulnerable to it if they don’t hold on to a sort of core “go touch grass” reality. Which is that it takes a long time to learn some things, and human judgment is valuable.

Warzel: Where do you all think we are in terms of adoption? Because again, for someone on the outside, who’s not speaking to managers and the rank-and-file types of employees at all times, it’s hard to get a good sense.

But how much have these tools already changed what is happening day in, day out? Versus how much of it is that feeling of like, I need to be doing this. This is something that we need to have? There is the, you know, the FOMO element. Where do you see the balance there?

Johnathan Nightingale: Certainly among the groups that we work with, people in engineering roles have been the most curious about it, and in some cases the most compelled to engage with it. You definitely have people who are very credulous—“We’re playing with it a lot”—who feel like they’re getting super productive about it. We’re starting to see studies about how those people are burning themselves out. Right? Those people are really struggling, because they’re orchestrating so many bots that they never want to close their laptop. Because they’re getting this dopamine hit from productivity. They’re not necessarily deepening their skills; they’re just doing a bunch of stuff. But like, there you see a ton of adoption.

Melissa Nightingale: We’re starting to see the fingerprints of AI-adoption mandates in context where it makes no damn sense. So like, prompt saying, “We have a client organization that was building technology to respond to bots infiltrating contact forms on websites.” Because so many people were having their agent basically go: “Reach out to this organization; we’re trying to get this work done. Can you go figure out a quote for this thing? Like, go write them with a description of what we’re trying to get done; have it come back.” But the cycle times are considerably longer, in part because the context that you can anticipate a human responding to it needing just isn’t there. And so you end up with, like—it’s meant to save a step, but it causes three more. And we’ve all had the experience of somebody like a junior person sending a shitty email and being like, Fuck, I gotta go unwind that. Right? Like, I gotta go unwind that you sent an email. And, it makes no damn sense. And now I’ve got 30 people in the organization going on, running and chasing a thing.

Johnathan Nightingale: Or “You sent it to a client, and now I gotta go apologize for that.”

Melissa Nightingale: Right. But imagine that multiplied across many workforces right now. Where you’ve got a lot of communication and requests flowing that, like, just missed an important key step before they went out the door. And so there’s a bunch of weird context and cleanup that’s taking longer than the original initial task would have taken to just do the damn thing.

Johnathan Nightingale: And there’s this rigidity forming, too, which is worth fighting against. Which is that you’ll have people farming everything out to GPT—even really obvious ways, right? You get emails from a colleague, and you’re like, “This is not what you sound like. This is obviously GPT. And what am I meant to do with the fact that you didn’t bother to write this email?” Right?

Warzel: Absolutely.

Johnathan Nightingale: And then like the alternative is to say, “Well, screw it. I will never let GPT do that. It’s an insult when somebody does it to me; I’m not gonna do that to other people.” And from that place, this sort of entrench and say, I refuse to engage with these tools. And then you’re putting your refusal up against management pressure. And, it’s driving conflict that isn’t very helpful.

Warzel: When I hear, “Oh, these agents are filling up these contact forms; it’s creating a whole bunch of extra work.” The guy who’s been listening to people in Silicon Valley talk about this technology for a very long time says, “Yes, well; their solution to that is you need agents on both ends, because having agents on just one end is, you know, not balanced.” There’s this way in which I can see that. People are creating—it’s a very classic tech thing to create a bunch of problems and then offer a technological solution to the problem that you created.

Johnathan Nightingale: Yeah.

Warzel: That all adds to more problems down the line. I also think, though, that what I wanna highlight there is the way in which this creates these asymmetries and builds these little cracks and pieces of distrust. That seems like something that feels really important in the context of all of this going forward, if you are somebody who cares a lot about what you do and puts a lot of effort into these things.

And then you have a group of people who may also care, but they use these tools. And there’s not this standardization of work output. And you see somebody respond to your email with something general. I think it’s really interesting that it creates that fracture, based off of how AI-pilled you are or how excited you are about the technology. And that feels like a bigger problem than I think people are thinking about right now.

Melissa Nightingale: We’re seeing that play out across like technology right now. Where you’ve got folks who feel like maybe you’re being incredibly efficient, and you’ve got like 12 agents working for you, and you’re getting a ton of stuff done. And I think, Wow, what a go-getter. Or I think, You’re really fucking rude, and you don’t give a shit about your work or the impact to my organization on you sending garbage across as a transmission.

Johnathan Nightingale: Isn’t that interesting? Like, not from a culture war, “red team versus blue team” shit. But like: why? Like, why can so many people nod along to this idea of like, Oh yeah, when you get a GPT email, that feels rude. Like, That feels like the person didn’t care. Right. I thought these agents were supposed to be solving all kinds of problems. I thought they were more talented. They’re passing the LSATs. They’re like, you know—they’re doctors, they’re therapists, they’re whatever. Like, why would we receive it as rude?

It turns out that humans really care about doing work they believe in, with people they care about. And when you hollow those things out, people have these emotional responses to it that I don’t see predicted by the marketing materials from the AI companies.

Melissa Nightingale: And the hyper-rationalists will fall short on this one every time. Because, fundamentally, being more efficient should, like, it goes—

Johnathan Nightingale: You ought.

Melissa Nightingale: You ought to be excited about being 10 times more productive than you were yesterday. You ought to be excited that your colleagues sent you an email that they didn’t spend any time on, because you also don’t have to spend any time on reading it or consuming it. Like, that should be exciting.

Warzel: You’re both management and leadership trainers. Which means your job is ostensibly to try to make people better bosses, right? And I wanna ground some of this in asking: What makes a good manager? Or what makes a good middle manager? What are those qualities there?

Melissa Nightingale: We have bosses who show up in programs. And they’re like, “I am a good manager because my team likes me.” Wrong. “I am a good manager because, like …

Johnathan and Melissa Nightingale: “My team is happy.”

Melissa Nightingale: We’re like, “Wrong.” Happy is an impermanent state, right?

Johnathan Nightingale: And you can’t take custody for people’s emotions.

Melissa Nightingale: No. Fundamentally, you are a good manager if you are making your team more effective.

Johnathan Nightingale: And effective feels really mercenary, but it isn’t. Because it turns out as you get deeper into this, you learn that the best way to build an effective team is to see them as individuals: to align their personal motivations and aspirations and sense of mastery, the things they want to learn, their curiosity with the things your organization is trying to get done. It turns out that, like, that only happens if there’s a high level of psychological safety. If they’re able to take risks; if they’re able to talk to you about their struggles.

Melissa Nightingale: If you’re able to give critical feedback on the work, where it is showing up exactly as it should and where it isn’t showing up exactly as it should.

Johnathan Nightingale: And if they see you engaging as an authentic leader. They don’t need you to be perfect, but they need you to not be bullshit, right? It’s an important part of how they cohere as a team, and how they find anything that you have to say remotely credible. But many people in management roles lack those skills. And if you’re like, “Wouldn’t that make work terrible for a lot of people?” Yes.

Warzel: There we go. Well, part of the reason I ask is because it seems in this AI conversation, the term management is a skill set that AI companies seem to want to impose on all of us, right? Recently the co-founder of Anthropic, Jack Clark, went on a podcast, he’s talking about the way that chatbots and increasingly these coding agents are, you know, going to be sent off to accomplish these tasks. And there’s going to be all of what we’re talking about, right? These things happening on behalf of us: goose chases, all kinds of stuff. And he had this quote that I thought was striking. It’s one of the reasons I wanted to have this conversation with you all. Which is, he says, quote, “Everyone becomes a manager, and the thing that is increasingly limited—or the thing that’s going to be the slowest part—is having good taste and intuitions about what to do next.” What do you make of that line? “Everyone becomes a manager.”

Johnathan Nightingale: It’s such a shallow read on management, isn’t it? Like, if you think about it—let’s say he’s right. Let’s say he’s right that when I fire up my Claude Code instance and I say, “Claude, create a marketing team for me. I want a content marketer. Want somebody on social media.” Right. Have I done it now? Like, can I transfer those skills over to the humans that are still on my team? Is it the same thing? Can I just go to Tony and be, like, “Tony, get Alex and Sam in a room? You got new jobs. Now. Here’s what you’re doing.”

Is that going to work? Nobody thinks that’s going to work.

Melissa Nightingale: I also think any model for management that doesn’t come with employees who have needs and, like, labor and rights—there’s a bunch of pieces that are missing from that mental model. And either that’s accidental, or that’s on purpose. But we should all be really concerned about a model of labor and a model of management that includes work happening without any capacity for what the people doing that work—or what the folks who are supposed to be responsible for that output—need.

Johnathan Nightingale: Buddy seems to know a lot about AI. I know the podcast you’re talking about. It’s cool. He’s built something that’s very big and seems to be changing a lot of people’s lives. And, I hope, some of them for the better. Like, that’s super neat. But on management, I’m not convinced he has the range to give anybody advice on what constitutes good management. You can build a very wealthy company and still be cooking a bunch of people inside it. And saying something like that—I don’t know, you can call that an insult, you can call that a threat, but that’s not what management is.

Warzel: Well, what’s interesting is, you know: As you’re describing what makes a good manager, and you pause and say, “This could sound mercenary.” Right? But what’s actually involved in all of this is the very human work of recognition. Of getting to know people. Of caring. Of giving a shit, right? It seems like that idea of “everyone becomes a manager”—that definition from an AI executive—it pauses at the mercenary point. It says, “He’s with you until that moment where you say, But then it requires all this.”

And part of the reason why I want to have this part of the conversation is also because my partner, Anne Helen Petersen, and I wrote this book about the rise of remote work in 2020. We spoke to you guys a lot about that book. And one of my broader takeaways from all the reporting is that—despite being in the boss layer, right?—the managers and especially middle managers were pretty miserable. Just in general. Just, like, a pretty miserable, core of that thing. Like, maybe miserable in different and unique ways than, you like oppressed rank-and-file workers. But pretty miserable. S

So when I hear that we’re all gonna become managers? From that, I hear: Man, that’s a lot of task switching and a lot of, “I’m getting incoming from two sides of this thing.” Like, two groups of people are kind of converging on me. I’ve got these unhappy people on one side; this unhappy thing on the other side. I kind of have to figure out how to exist in this world. I’m being pulled in a hundred directions.

Melissa Nightingale: Charlie, you’re selling it. You’re making it sound so compelling.

Johnathan Nightingale: Strong pitch, strong pitch.

Warzel: But that, to me, is like: I don’t know. I mean, does it seem as dystopian to you? Or do you just recognize this as, I don’t know, a guy pontificating? Without, as you said, the understanding?

Melissa Nightingale: We have a working mental model; like, we have a real-world model for where automation takes more of the front seat on management. Amazon warehouses are a good example. Uber drivers are a good example.

Johnathan Nightingale: Already being managed by algorithms. The common thread there is this idea that where people are working in Amazon warehouses because they can’t automate that piece yet. Right. And so they build all the automation around these little stations where the human stands, and the humans are reaching up and reaching down. That shouldn’t be our vision for the future of work. And when you listen to a lot of the people who are like, “Oh, everyone’s going to be a manager.” You’re to have this swarm of agents. Who’s doing stuff? Like, what’s the end state there? That you’re a team of one? That you’re a company of one? And so, when you try and sell me this story about a billion-dollar company with one person, I’m like: Think about the best times you’ve had at work. Best times. You might have had a shit boss. I get it. There’s a lot of them. We’re working as hard as we can. But, like, the best times you’ve had at work—and I guarantee that story involves colleagues. Right? And so when somebody tries to sell you on, “Don’t worry; you don’t need colleagues anymore,” I’m like: What are you doing? Like, that’s an anti-signal. I understand your technology is very impressive, but, like, that’s a weird thing to sell.

Melissa Nightingale: Unless you’re annoyed at paying the junior engineers $300,000 a year straight out of school. And then it’s a very compelling sell.

Warzel: There’s a lot here, right? Some of these people who are talking this … some of them, yes, are people who are inside large companies right now that are building all this. That have this workforce. So I think that argument makes a lot of sense. There’s another group, though, of, let’s just say “venture capitalists” who are actually on that island, right? A little bit. I mean, yes; there are plenty of people who work in some of these venture firms, no doubt. But the idea is that sort of like, life of the mind. Like “I’m this tech soothsayer.” And I think in some sense there is this, “How can I push this to the furthest possible extent?” Right? And there’s this idea, always, with the efficiency, right? The reason why the AI bro–type person who’s saying this thing about the excitement of the first one-person, you know, unicorn company, I think, speaks to the idea of … like, this is “efficiency” pushed to its broadest level, right? This is like the cheat code for late-stage capitalism.

And I think it ties though to this broader premise of how productive this stuff actually is, right? So there’s this recent survey from this company called ActiveTrack. And they analyzed over 10,000 workers across 376 companies. And they did it 180 days before and after AI adoption. And the thing that I think won’t be surprising to a lot of people who follow this is: email, up 104 percent. Chat and messaging, up 145 percent. Collaboration time surged 34 percent to an hour a day. Multitasking rose 12 percent. It’s that classic, you know, Parkinson’s-law, “work expands to fill the time available for its completion”–type thing. What do you make of those types of numbers and the idea of productivity?

Melissa Nightingale: We tell bosses all the time: Busy is not the same as effective. Like, you can run your team totally ragged in terms of having them chase every idea that seems like a good idea. But if that’s not what we’re here to do, then you’re wasting their time and your own.

Johnathan Nightingale: Yeah, the maximalist arguments are always so weird, right? Like, the thing you could do is say to your team: “This is a cool tool; we should figure out where we can apply it.”

Melissa Nightingale: Where is there an annoying problem within the organization that you wish someone would fix for you? Where is there an internal tooling thing that would be so cool to have and get a thing out of your way in your workflow? Like, great—let’s go build that.

Warzel: Right. Software-size problems.

Johnathan Nightingale: That could be the end of a sentence. It doesn’t have to be. And then we fire everyone. Like, it doesn’t. You can just end. They’re like, “Well, but if all your competitors are cutting to the bone and outsourcing all their sort of, you know, friendships to ChatGPT or whatever, then you’re going to have margin pressure. And you’re going to have to do the same thing.” And I’m like, man, “Maybe.” But it turns out that creative, resourceful, adaptable humans are good at some shit. And that an LLM—which however amazing you might find it to be—is trained on yesterday. It’s at some point going to run into problems inventing tomorrow, right? And like, that’s a thing people can do. And then tomorrow it’ll train on it. But I will bet on people. Again, not like a team thing. It’s cool for the people to have tools. “We’ve got lots of tools.” That’s great.

But it’s so weird to bet your business on entirely outsourcing critical thought, creativity, collaboration, partnership to this thing that can generate grammatically correct paragraphs. Like, it just feels like such a weak-sauce version of leadership.

Warzel: How hard has it been to give that message to this group of people? Have you been effective in conveying that? Is the siren song of all of this, you know, too difficult to resist? Like, what are you butting up against with this?

Melissa Nightingale: Two things. Okay, so it’s a great question. Like, on the one hand, you spend a third of your waking hours at work. If you’re lucky and have like a normal work schedule, you spend a third of your waking hours at work. So I think for a large number of people, the idea that like one—we just fully and outright reject the idea that that’s how we’re going to spend a third of our waking hours, as a collective. Like, absolutely not.

Two, we’ve seen it, and that lends some credibility. In that we’ve seen inside a lot of organizations, and we talk to a lot of leaders about moments at work that really mattered to them. And it’s like: I think the conventional wisdom is nobody would have any moments at work that mattered to them. Nobody would have any moments where a boss saw them, connected with them, brought them up, helped skill them up. Helped unlock the next stage of their career, because work is crappy. We have to spend a third of our life there, and it’s always going to be crappy. But the fact of the matter is, you ask people: Do you have one of these moments? Do you have one of these things that happened where work really mattered to you, or was a support or a stability for you at a time where the rest of your life was, like, really rocky? People, by and large, do. And so our starting point, I think, for a lot of folks is that the idea that it could be good for a lot of folks is one, a radical concept, and two, a very welcome idea.

Johnathan Nightingale: But when you ask about receptiveness, it’s funny. One of the first companies we ever worked with was an AI company. And I remember meeting with the leaders that were going to be coming through a program with us. And this one guy, very sort of eccentric-professor vibes when he came in. Sort of scattered, came in a couple of minutes late. And he said, you know, “I’m coming to this management program. They’re sending everybody to this management program. But I need you to know something right up front. I don’t believe any human should manage any other human.”

Warzel: Yeah.

Johnathan Nightingale: Cool, fair, yeah, great. He’s a bit, “I’ve never done a program like this before; so you know, I’m open to it. I just want you to know that upfront.” By like the third week of the program, he’s sitting there reading High Output Management by Andy Grove. Which is sort of, former CEO of Intel; like a standard management text, particularly in tech circles. We’re talking about hiring, and he’s looking at a job description. And he’s like, there’s a bunch of gendered stuff in that job description. You’re going to end up with a really tilted candidate pool if you keep doing it. Like, he’s fully engaged with it, and is really thoughtful and conscientious about how to engage with it. He just didn’t know the receptiveness is a really easy sell. It’s a surprisingly easy sell. Nobody likes to be in a job that they don’t know how to do and feel like they’re failing all the time. And if you can give them some stuff, you know, there’s this moment where you give them some tools. And they’re like, “I don’t know if that’s going to work.” And then you get them back next week, and they’re like, “That did work. What else do you have?” Right. And like, it’s actually really easy to convince.

Warzel: How worried, just hearing you say this, how worried are you guys that these tools, these AI tools, will effectively just act as a Band-Aid for any of these moments? Of “Instead of having to think about it hard, and have that conversation that’s ultimately really fruitful and ultimately helps me and everyone else around, I’ll press this button instead”?

Melissa Nightingale: We come back around again, right? It’s like, if I am your direct report and I fucked up in a meeting, like: I’m sorry, I just didn’t know that thing, and I said that thing, and I thought it was a good idea at the time. But like, whatever, I screw up in a meeting. And what happens after that meeting is you send me a thing to tell me I screwed up in that meeting, and it is perfectly outlined.

Johnathan Nightingale: Lots of em dashes.

Melissa Nightingale: It is perfectly structured. And, you know, you’ve given the feedback to an LLM, but it spit out a thing. And then I get the thing; we’re back to the part where I’m like, That’s rude as shit. Like, if I screw up, tell me. I am a grown-up; I can absolutely handle it.

But you’re back to this weird moment in work culture right now. Well, either that’s a well-intentioned manager trying to format a thing so that it lands correctly without having to, like, have the like social risk in that moment. Of, like, “If I send it to Melissa and she’s upset about it … well, if GPT wrote it, she’s mad at GPT. She’s not mad at me. But if I actually put time and effort and energy into it, I might get it wrong. It might not land well. And I might have to spend some time reflecting on, like, That isn’t the way I wanted that to go. I want it to go differently next time.”

Johnathan Nightingale: There’s this study. Even if it never touches the employee, even if it’s just something that the manager does, you know, they go into their own chat window and say, “Here’s what happened. What am I supposed to think about it?” Study came out a little while ago, talking about sycophancy in LLM chats and its effect on sort of attribution of fault during conflict. Right?

So I get into a fight with someone, or my direct report is pissed off because they didn’t get a promotion. Or I’ve got hard feedback to give to them, and I gave them the feedback, but it didn’t go the way I wanted it to. And so I’m using Chat to help me debug it. And what the study found is that the more sycophantic the LLM is, the more likely I am to feel like I did nothing wrong. The more likely I am to bring that sense that I did nothing wrong into my future interactions with that person. And then you cross-multiply with the studies that are well established now, that LLM use suppresses critical thought and critical reflection. And that a major component for leadership development is critical thought and critical reflection. And you’re in a bad spot. Even if the employee never receives that GPT message, just me using GPT as my leadership coach is likely to really impair my sense of my relationship with my people, and also my own reflection.

Melissa Nightingale: Outside of management and leadership circles, it may not be obvious why that’s such a core component. But basically: If you are in a modern organization today, one of the biggest problems that we have of getting bosses to show up differently in the role is that they are often running from meeting to meeting to meeting to meeting to meeting. So if something goes wrong in my first meeting of the day, I have no opportunity—other than 2 a.m. when I’m staring at the ceiling—to think about, like, How did that go, and what do I want to have happen next time? Learning like humans are really freaking good at like: “This went this way; like I touched the hot stove. I don’t want to touch the hot stove again. And here’s what I’m going to do differently.”

If they’ve got time to think about it. The hard part that we have, in our industry and in our sort of line of work, is that for many leaders there’s not any of that baked in anymore.

Warzel: I want to get here, near the end, of something bigger, though. That we’re kind of circling around all this. Which is: this idea of a constant replacement in work, right? Replacing the expectations of how much work should take up someone’s life. You know, the expectations of how much dignity someone should be able to derive from it; the expectations of the path of predictable progress in a given career. The technology is often that we put into these spaces, you know, as we said. They free up space that is then used to, you know, fill more in. You keep piling stuff on here, without the attention and as much of the focus on the nourishing qualities that we’re talking about here. And the things that might give people some purchase and some connection. And you’ve all been writing a little bit around this idea lately of social atrophy, and the role that work plays. What is work’s role right now in this broader loneliness epidemic? Third-space dwindling, broader-disconnection feeling that a lot of people are experiencing?

Melissa Nightingale: We started to see weird glimmers about two years ago. Where a lot of organizations were saying, like: “I’m managing a team, but my team is geo-distributed, right. And so, I’m managing people all over the world. And sometimes I’m managing them, occasionally in office, but sometimes I’m managing them [remotely] … and we just don’t share a lot of overlapping daylight hours. And we work in sort of our own Zoom windows. And for my folks, where I’m managing them, and they’re not interacting with a lot of people, things have gone a little bit weird. And what do I do about that?”

Like as a management question, right? Would come into our programs and say, like, “It’s not a performance problem, per se. They’re showing up for meetings. Like, the camera’s off, but they’re showing up. Things have just gotten wobbly enough that I have an overarching wellness concern. What do I do with that?”

Johnathan Nightingale: And the more we looked at it, the more we saw … you know, Melissa uses this language of work as the last bastion. Robert Putnam wrote Bowling Alone, right, and talked about “People aren’t in bowling leagues anymore, but they also aren’t in rotary clubs, and they also aren’t in churches.” That whole sense of, like, our community glue is, you know, eroding. If you want to take the worst version of it. Or certainly evolving.

But through all of it—work is a place that you show up, and you’re around other people. And, you know, they see you and appreciate you when you do good things. And give you interesting things to work on. Or at least give you interesting things to talk about while you’re getting coffee. When that falls apart, there isn’t another backstop. There isn’t another place we spend eight hours a day, five days a week.

Warzel: So you feel like these tools, as they’re further embedded in the workforce—especially if they’re embedded, you know, not critically—are a real threat to that.

Melissa Nightingale: I think if we had community answers for human connection, I would be less worried. But we have eroded most of our social answers that are baked in. We don’t live near our relatives anymore. Most of us sort of go away. And then we sort of set up new communities. And maybe we’ve got some chosen family. But we have a really different context. And our backstop—for a lot of us, sort of last bastion—was work. Where we had to go, and we had to socialize. And even when we didn’t feel like it, we had to, like, put on our hard pants and get ourselves sorted. And, you know, do the thing. And humans are social creatures. Like, it is core to who we are, the whole way along.

Warzel: I fully agree with that. And yet I wanna push back slightly, because I can imagine there’s people listening here who are going to say: “Work is work. Work is not family; work should be transactional.”

Johnathan and Melissa Nightingale: Totally. Yeah.

Warzel: And these AI tools depersonalizing work in some way—in making it so that like, yeah, when Tony sends that email, and it’s very clear that he doesn’t care, because it’s a chatbot—ultimately, that may be even a better thing, right? Or these tools are going to free people up to not have that time. I think we’ve sort of debunked that part of it.

But this idea that impersonalization is actually a feature, not a bug, right? That these companies … we were talking earlier about Amazon warehouses and things like that. One of the rebuttals to that from the AI-evangelist guy on X.com is gonna be like, “Who cares, man? It’s a business. We’re supposed to wring out every inch of productivity, or whatever. And yeah, there’ll be more bots in the chain, so that we don’t have to hang up ibuprofen because people have repetitive-stress injuries.”

This is sort of the true mercenary level. But also the mercenary level, I think, on the sense of employees who are like, “I’ve had it. I’ve been exploited by the system for so long. I actually like the idea that this is going to feel depersonalized.” How do you see that butting up against the last-bastion status?

Melissa Nightingale: We meet a lot of those folks. We meet a lot of folks who are like, “I just don’t care anymore. I have cared a lot. And I really feel like I’m ready to care less.”

It’s hard. And the reality for those folks is … it’s tricky. You can put yourself into a role in an organization. That you think you’ll be fine. And like, you will still manage to get yourself promoted. And they will still figure out that you’ve got ideas, and you will still find your way to … like, they often find their way back to, “Okay, I do care about this. I don’t necessarily need it to be like my entire waking hours, or my entire personality. I need some space from it, and some other elements in play.”

But it’s hard. It’s very hard. A lot of folks have work as a key component, and a core element, of their identity. And you can say, “Well, they shouldn’t. Like, they’re silly; they’re foolish for feeling like that’s a moment of profound identity shift.” But again, we have to work with humans as they are, and not as we wish them to be. I think it’s okay that people care about their work.

Johnathan Nightingale: Yeah, when you hear that—have boundaries. Get out of toxic workplaces, right? Because nobody’s saying “Wherever you are, that’s where you must be forever. Otherwise, you’re ungrateful. Otherwise, you’re not applying yourself.” Like, that’s foolish. Like, if you need to get out, get out. If you need to keep yourself safe, keep yourself safe. But, in a deeper sense, I don’t believe you. Like, we do actually care about this shit, and so, like, if you don’t care about your job, that’s fair. Do whatever you need to do. Capitalism’s mean, right? But, you will not sell me on “Nobody cares about their jobs” or that it’s not worth caring about your job. Because so many of the people we find who draw so much meaning from it don’t see themselves in that at all.

Warzel: So to end then, with that in mind, how does that intersect with this idea that, you know, these AI tools are going to destroy knowledge work or white-collar work or whatever? Right? Like, is that a bulwark against it? The idea that there’s a lot of people out there that fundamentally give a shit? Like, ’cause it feels to me, if that part is really true, you’re going to have this technology come right up against people, like a major pillar of who they are, right? And I don’t think that there is a sense that yes, capitalism can sort of just, you know, knock people over, bend people to their whims. But I also think that, it seems to me, like this is all geared to meet some really heavy resistance from people who work. Do you think that’s true?

Melissa Nightingale: Remember that protest looks a lot of different ways. But like, me feeling like, as an employee, I’ve got options in terms of where I work. And who my colleagues are, and whether I have colleagues at all. Means that organizations that sort of lend themselves to that—or sort of specifically go out with a message that says, like: This is what we’re doing. We’re gonna have colleagues. It’s gonna be great. You’re gonna love it. Like you’re gonna be in an occasional meeting that isn’t useful. It’ll be okay.

Johnathan Nightingale: But there’ll be small talk beforehand. And nobody likes small talk. But it is interesting to see, you know, how her vacation went and stuff.

Melissa Nightingale: But like, you will see change, I think. Less through walkouts and more through people feeling like the pendulum swings back, and organizations are trying to hire again. And AI companies, in particular, are very skilled at what it looks like to have a very hot talent market and to have to compete on the merits of the organization.

Johnathan Nightingale: Isn’t that wild? That while they’re telling everybody else to lay off your team, and pay the remaining ones as little as possible for doing 10 people’s worth of work.

Melissa Nightingale: They are having 2021’s version of the labor market.

Johnathan Nightingale: Like, just throw any amount of money at getting top talent. We need the right people in the door, otherwise we’re not going to be able to build the future. Like, what? I thought Claude was doing that.

Warzel: Melissa and Johnathan, thank you so much for coming on Galaxy Brain and talking about the weird future of what we all do.

Johnathan Nightingale: Anytime.

Melissa Nightingale: This was a lot of fun. Thank you.

Johnathan Nightingale: Thank you.

[Music]

Warzel: That’s it for us here. Thank you again to my guests, Johnathan and Melissa Nightingale. If you liked what you saw, new episodes of Galaxy Brain drop every Friday. You can subscribe on the Atlantic YouTube channel, or on Apple or Spotify or wherever it is that you get your podcasts. And if you wanna support this work and the work of my colleagues, you can subscribe to the publication at TheAtlantic.com/Listener. That’s TheAtlantic.com/Listener. Thanks so much, and I’ll see you on the internet.

What Is Twitter’s Legacy, 20 Years Later?

2026-03-27T13:00:00-04:00

Subscribe here: Apple Podcasts | Spotify | YouTube

What is Twitter’s legacy? In this episode of Galaxy Brain, Charlie Warzel traces how Twitter, now called X, evolved from a status-update tool to one of the most culturally and politically influential—and contentious—platforms of the modern internet. Charlie is joined by early Twitter executive Jason Goldman. They explore how Twitter’s core features—many invented by users—reshaped media and politics while also enabling new forms of harassment, misinformation, and attention hijacking.

Goldman reflects candidly on the company’s key inflection points—from early free-speech-maximalist decisions and underinvestment in trust and safety to Twitter’s role in events like the Arab Spring and the election of Donald Trump. The discussion culminates in Twitter’s Elon Musk era, where its logic of attention has been weaponized more explicitly. The episode reckons with what Goldman and others ultimately built: a tool with outsize cultural influence that’s broken brains and amplified some of society’s worst impulses.

The following is a transcript of the episode:

Jason Goldman: When you’re that zealous about the mission that you’re on, which you almost need to be in a startup to survive, that zealotry blinds you completely to the downside risks that you’re producing.

[Music]

Charlie Warzel: I’m Charlie Warzel, and this is Galaxy Brain: a show where today we are going to try to explain how Twitter, this niche microblogging site, became one the most influential social networks ever and broke countless brains in the process, including my own.

Last week marked the 20th anniversary of the first tweet sent by one of its founders, Jack Dorsey. The post, which read “just setting up my Twitter,” is a pretty good example of how far the platform has come. By July 2006, Twitter was available to the public. And at the outset, Twitter was about these little status updates, a lot like the old AOL instant messages or away messages. People posted ambiently about what they were doing. And this was all pre-smartphone. It was this silly, low-stakes, localized way to communicate. But like a lot of technologies, it took off, became something almost totally unrecognizable.

My own relationship with Twitter, like a lot of journalists’, was always really tortured. In its early days, the platform was this way that young writers, reporters—they all got noticed. And to this day, I argue that I owe certain parts of my early career to early Twitter. And yet, I’ve spent the last 15 years covering its noxious effect on women, people of color, and bystanders who’ve had to endure intense trolling. The effects on our politics and our culture.

Twitter’s founders and leaders espoused this free-speech-maximalist approach, which didn’t just allow for abuse and hate speech; it helped build a platform that optimized it.

Now, you cannot tell the story of the last two decades without Twitter. The rise of Donald Trump, the mainstreaming of this trolly message-board culture, attention hijacking, the real-time radicalization of politicians and tech elites. And of course Elon Musk: the centibillionaire power user who bought the site in 2022, fired the bulk of its staff, and has since turned it into his own political weapon. Twitter, now X, is quite different.

Its worst qualities are on full display now—not as bugs of the platform, but as features. And in some spheres, especially Silicon Valley and the AI conversation, it still manages to have an outsize influence on broader discourse. For whatever reason, the platform seems unkillable.

And so, what is Twitter’s legacy? Why did things turn out this way? How did we get here? To try to get to the bottom of this, I asked Jason Goldman to join me. Jason is one of Twitter’s earliest employees—joining the company from Google, where he worked on Blogger, in 2007, and rose to the VP of product at Twitter. Goldman was influential in shaping parts of the platform, including its controversial content-moderation policies.

He was in the room for most of the platform’s early successes and its failures. And he’s somebody who’s had to watch the way the platform that he once stewarded has warped and weirded society from the outside. He was fired from Twitter in 2010 and later went on to become the White House’s first chief digital officer under Barack Obama. So together we trace the history of Twitter from its hackathon foundings to Musk’s takeover. What made Twitter special, its original sin, Goldman and the founders’ biggest and most costly mistakes, and why the platform, 20 years later, is still alive. He joins me now.

[Music]

Warzel: Jason, welcome to Galaxy Brain.

Goldman: Thanks so much for having me.

Warzel: Absolutely, I’m thrilled you’re here. 20 years of Twitter. As we’re recording this, a couple days ago was the anniversary of the first tweet. You were there at Twitter about as close to the beginning as one can be. And so I wanna start with something I think a lot of people here probably just don’t know, which is that Twitter was, before it was Twitter, a podcasting company or a podcasting idea.

Goldman: Podcast platform. Yeah.

Warzel: Yes, yes, a platform. Tell me about the early days. What is Odeo? How did it become Twitter? Take me back.

Goldman: Yeah, so I think from my perspective—and this is, you know, influenced obviously by my own history—I think you have to go back to the Web 1.0 days of blogging, which everyone loves to talk about. And so I worked on a product called Blogger that was acquired by Google in 2002; end of 2002, beginning of 2003. There were six of us. Three of us were named Jason. And all six of us went to go work at Google in 2003. Ev Williams was the CEO of Blogger, and one of the first people we hired at Google was Biz Stone. So two out of the three co-founders of Twitter come from Blogger. And one of the things that we worked on while we were at Google was this product called Audio Blogger, which was a partnership between us and this guy Noah Glass, who ran this third-party service for posting audio snippets to the web.

And then Noah, Biz, and Ev went off to do Odeo. They were interested in this idea, of There’s something here that’s interesting about posting audio to the web. Let’s make a podcasting platform. Let’s make the YouTube of podcasts in 2006. I was not very interested in podcasts; thought podcasts seemed not like a really cool idea. And so I stayed at Google while they were working on Odeo. And Odeo, you know, kind of had an interesting notion of what to do, but it was just way too early for podcasts.

Odeo kind of went sideways, and they had this idea of doing this hackathon where a bunch of different people at the company would break into teams and come up with different ideas for things that they could try, most of which were in the kind of social space. And then Jack [Dorsey], who was an engineer on Odeo, and Noah and Biz and Crystal [Taylor] were sort of the four people who worked on the team for what became Twitter. And it was an instant success internally. It was something like, from March 2006 it was clear there was a lot of interest in it. And yeah, was something that there was clearly legs, from the very beginning.

Warzel: So you’re working on Blogger. Obviously, I mean, if we think about it now, such a fundamental piece of media technology. We still use the name. What makes you jump over in that really, you know, exciting, energizing moment for Blogger?

Goldman: There’s two things. One was: I’d been at Blogger for almost four years. And so I’d seen I’d been able to do kind of a lot of different things there. But there I had sort of hit the bulwarks of the larger Google organization pretty hard. And Blogger was just never a good cultural fit with the rest of Google proper.

And then two: In 2004 to 2006, at least the founders and sort of the executive team of the product-management side of the engineering side, really did not get what blogging was. Like, all of the things you’re saying about it being like the “heyday of blogging,” being cool. They did not care about those things at all. And, fundamentally, we would have conversations with like, you know, Larry [Page] and Sergey [Brin], and you know, the rest of sort of the executive team there. About like, hey, like, “You know, we’ve got more pages than The New York Times; like we’re a huge site on the internet.” And they’re like, “Yeah, but The New York Times is where you go to get to the news.” Like, when are you going to have something that’s authoritative that people can trust? Which is not an unreasonable question. But it sort of misapprehends like what the point of all of this was. The cultural influence of blogging was not something that was well embraced by the company that we were working for. And that was, like, ultimately kind of felt annoying after a while.

Warzel: And so you’re feeling this. You are talking, I assume, to Ev and Biz.

Goldman: Ev and Biz. Yeah.

Warzel: Is that the thing that drives you over? Or is it more the broader umbrella of “These guys are smart. They’re trying to build a bunch of different things.” Is it Twitter that attracts you? Or is it—

Goldman: It’s Twitter. It’s Twitter, for sure. Like, you know, Twitter—the first tweets are from March of 2006. My account starts in May of 2006. I’m on a trip with Ev, and he’s like, “You should check out this thing.” I signed up over SMS; never see the website. Ev actually creates my username, and I was good. We had three Jasons at the company at Blogger. So I was always called Goldman, which is why my username is Goldman as opposed to … I could have had Jason. It’s just like, Ev created it for me. And I had this camping trip with Biz and Ev that summer. I was like, “How do I work on this? I want, you know, I want to work on this thing.” And they’re like, “Well, you know, we don’t really know what it means to work on this.” Because like, you know, “We’re doing this other company, and it’s not really working. This is a side project of that. And we’ve got, you know, there’s a product manager we already have there.” I was like, “All right; well, I’m going to quit my job at Google. I’ll travel for a couple of months, and hopefully by the time I’m done doing that, you’ll have cleaned all this up.” And during that summer, fall of 2006, Ev buys back Odeo from the investors, and it creates the Obvious Corporation. Of which Twitter was meant to be one of many products that would be worked on. And I’m hired as the director of product strategy for Obvious.

Warzel: Okay, what was the thing on the trip that caused the spark for you? If you have to drill down, what was the thing about the service at that time that did that?

Goldman: There were six of us. Went to Vegas, and we all had like sort of slightly different interests in Vegas. My interest in Vegas has always been very middle-aged man. I like to have a nice dinner and then go play some poker and go to bed at a reasonable hour. Even when I was in my 30s, and the ability to use Twitter to like, just see what your friends were doing. Like I would just be like, “I’m gonna play poker for a little bit.” And it wasn’t like a text update to the group thread. And I would just be able to get this, like, sort of ambient awareness of what my other friends were doing. It’s like, someone’s going out to the pool. And there’s like six of us. It was like—it felt like I think, for people who like to have a little bit of social distance and don’t feel an obligation to kind of, like, have to respond right away to “Okay, I’m coming to the pool, too.” It was just like, I know what people are doing. I don’t need to respond right away. Everyone else knows where I am. We can coordinate, and things will emerge from this. Felt like a new way of being with your friends, both online and offline, that was simply not possible before that.

Warzel: There’s some big moments in the early years. You guys win, I think, best startup at South by Southwest.

Goldman: Best blog.

Warzel: “Best blog”? That’s what you won?

Goldman: A really weird historical fact was that we go to South by Southwest in 2007, which is widely pointed to as, like, Twitter’s “coming-out party.” We weren’t even on a panel. Like, we weren’t even, you know, Ev’s doing, or Jack is doing, like, a keynote to explain what Twitter is. Like, we weren’t famous enough yet for that. We were in the hallway outside in the conference center and set up a monitor, set up a screen where you could see all the tweets that were from people at South by Southwest. But the monitor of being able to see, like, there’s all these tweets. And you could post, and you’ll see your tweet. And you’ll be able to see, like, other people’s tweets, of what they’re doing. Created as, as Biz would describe, this opportunity for emergent behavior. Where you’d be at a bar, and all of a sudden you’d see a tweet from someone. Like, you know, actually there’s, something cool happening at this bar, like, six doors down. And you’d watch everyone walk out the door of that, and move to the bar six doors down.

Warzel: What were some of those emergent behaviors to, you know, quote Biz, like that came out of it? Where you guys were like, “Whoa, this is even more than the thing that we designed it for”?

Goldman: Yeah, there’s a couple different things. I mean, of course, the important context for 2008–2009 is that the service, like, took off. And was starting to, you know, was hockey-sticking. It wasn’t consistent. People think, “Oh, it just, like, looked like the usage chart for Anthropic” or something like that. Like, that is not true. Like, there were many periods in time where growth flattened out until some, like, new cohort got it. It was not consistent growth. So one of the questions I always get asked is, like, “When did you know it was working?” And I worked there until 2010. And through 2010, we consistently thought this could go away at any time. Like, At any time people could just be tired of this, and this could just completely die. So that’s one piece of context. The other is that the service just fundamentally didn’t work. Again, remembering the era that we were in.

There’s two major revolutions, paradigm shifts, that happened in this period of time in the late 2000s. One is the introduction of the iPhone, which is tremendously well-timed for Twitter and felt like a big breakthrough. The other one is the cloud-computing revolution, which we were not well-timed for. We were too early for that. So it just it simply fell over a bunch during 2008.

Warzel: Yeah. And that’s important, because this thing is having this increased cultural resonance and also developing this reputation for really being good when things are happening. Like during live moments, right? Like, you want to be a spectator.

Goldman: Yeah. Yes. Steve Jobs keynotes; Super Bowls. Like, those were things early on we saw as, like, people really like using it during these live events, and it would fall over completely during those events.

Warzel: Right. And you’d get the canonical fail whale, right? Which became its own thing.

Goldman: Yeah. It failed so much the failure became, like, an icon, like a mascot of the product. Which is I think, in miniature, sort of a pretty good shorthand for Twitter. Which is like: It is a company and a product that has failed so much that its failure is one of its most iconic mascots and images. And yet, it still existed.

That’s my context setting for 2008, 2009. You asked about, like, what were the breakout moments. I think a few, for me, the ones that felt most validating were the ones that I cared about personally. So it would be things like, my background’s in astronomy. So NASA being an early adapter of Twitter was really meaningful, because they both thought it was cool and wanted to use it to update about their missions. And came up with this genius invention, which was to tweet in the first person from the perspective of the probe going to Mars. And so the probe going to Mars would tweet, you know, I’m on my way. Like, you know, My shoots are deployed. And it felt like a live event that you were very personally connected to. And as a space nerd, it felt like the best possible use case. So that, to me, is always kind of No. 1 on the song sheet for me. I felt that was super meaningful.

Other ones include, certainly, the 2008 election. Where politics, it became clear, was going to be a dominant use case of the product. Those people, like, talking about news. The Obama campaign was using it to do engagement online in 2007, 2008, and we built a whole election site. And then, in 2009, after Obama was elected, we get called by the State Department because—this is timely—we are being told that Twitter is of use in Iran during these pro-democracy protests that are happening. And we have to stay up and not take downtime in order to help support the protesters. This turns out both, I think later, not to be particularly true. Like, it seems maybe Twitter wasn’t that important. But, and I think—at the time we internally did not want to make a big deal, or want to claim a lot of responsibility for pro-democracy movements around the world. Because we didn’t feel we understood those properly. But it became part of the media narrative about Twitter.

Warzel: Was it just a complete, like, shit moment when you get that? Like, We are in over our heads all of a sudden?

Goldman: Yes. We’re in over our heads. We don’t really know. There’s no adult in the room who’s like an expert. Like, in a contemporary company, you would at least have on the board, like, someone who came from a government background. Both Biz and I had a very strong read on that situation, which was the story is going to be that we’re being asked to like save Iran, for democracy. And we need to resist that narrative as much as possible. Both because we do not … we like the Obama administration, but we do not want to be an extension of the United States government. Like, one: We’re a private company. And two: We do not understand the particulars on the ground in Iran such that we can say whether or not what we’re doing is helping or hurting or anything else. Like this—all we know is that this is a complex, volatile situation in which we are not experts. And like, I take a lot of pride in that. Because I know for sure that it is not retroactively imposing a narrative; [we] have lot of documents that we both publish publicly as well as emails that we had internally.

I credit Biz a lot with the desire to push back on this. What you see a lot in tech right now, which is like: Any bro who has a Twitter account and a hundred thousand followers thinks that he’s an expert in international-energy economics. Or, you know, nuclear weapons, or you know. Like, everyone’s an expert in all of this shit. Like, we were not. We did not pretend to put ourselves out there as experts on the geopolitics of West Asia.

Warzel: Is this the moment where there’s this kind of a free-speech-maximalist ethos that starts to develop inside? Right? Which is this, like, “Hey, listen; we cannot, as we start to get more important, we cannot make these big, sort of hard leans on the editorial controls.” And that ends up being its own issue that Twitter has to deal with down the line. But is that sort of the genesis of it?

Goldman: One hundred percent. I think it predates that. I think it becomes tested in that moment, but the free-speech ethos for Twitter is a Blogger artifact, 100 percent. I can speak pretty definitively about this, because I was in charge of content policy for Blogger. And Blogger, we had a lot of arguments internally that went up to the executive team about wanting to maintain separate content policies for our Blogger. Separate from what were being applied to, in particular, AdWords advertisers. And we fought very hard against that, and ultimately won that argument and preserved a much more permissive view of what should be allowed. Which is, essentially, like: Unless it’s illegal, we should be very skeptical about taking it down. So then when I got to Twitter, before it was even Twitter, I wrote the first content policy for Twitter. They previously were using, in 2006, Flickr’s content policy, and Flickr’s content policy was very granola. And I was like, all right, we’ll have a few more things than that.

And we had some early tests of the content policy in Twitter in 2007, 2008, where people were being harassed on Twitter. And there was lot of—you know, there were like cases of someone being, you know, called names, and someone being kind of stalked across Twitter. Of saying, like, Oh, I know what you did with my ex. Or something like that. And we took a very hands-off approach to that. I think that’s … probably, I think we can call that a mistake. Think I made a mistake on that.

Looking back, we should have taken a more aggressive stance. I think the thing that we did not recognize until much, much later on Twitter was that Twitter was fundamentally a different product, because of the follow graph and the notifications about mentions. Once those became built into the product, it became much easier to engage in types of abuse vectors and harassment that was not possible on Blogger. Because on Blogger, you had your own little protected blog. You could just ban people from the comments or whatever. Whereas on Twitter, someone could show up in your mentions tab and actually be talking all kinds of terrible stuff to you. And even if you block them, you knew other people were seeing it.

So we applied this free-speech-maximalist idea from Blogger and kept it for quite a long time at Twitter. I think mistakenly. I think that was a mistake that I had a pretty instrumental role in playing. But it was because we did not recognize that these new kind of vectors were possible.

Warzel: I’ve done a lot of reporting around Twitter and online harassment and things like that, especially early on in those days. It was real.

Goldman: Yeah, 100 percent.

Warzel: There were people who suffered very real consequences due to this non-understanding. Like, without taking away anything from the seriousness of that or absolving anyone in this thing. It is, again, speaking to this era of whatever we’ll call it, now of Web 2—where it’s like “building the plane as everyone’s flying it.” Listening to you say that, it’s like you guys didn’t seem to really understand some of these emergence behaviors of the platform, because you didn’t know it until you saw them.

Goldman: Yeah. I mean, and to put a underline on that point, one of the primary vectors for abuse and harassment became at [@] notifications, right? You could show up in someone’s notifications and, you know, be talking really heinous stuff to them. The entire at mentions, like protocol, was user-created, right? That was something that users just started doing. It wasn’t a feature that we built. We later, then, built features to like kind of pave the path around that thing that people had started doing. But it was something that just emerged from how people were using the product. So it’s not even the case that we built—like, we kind of had this idea for how at notification should work, and how a notifications tab should work. And we didn’t kind of do the due diligence to think through the abuse vectors. It’s like: People just started doing this behavior, and we kind of put up a scaffolding around it. And then realized, Oh wait; there’s all this other stuff going on around the boundaries of that.

Warzel: And this—for people who aren’t intimately familiar with the history of Twitter or weren’t there around the early days—this is yet another thing that makes Twitter so much different, I think, than almost any other platform that exists. Which is that so many of these now-canonical features were just built by users. Like the hashtag, What I think is so interesting—and this has always been very interesting to me about Twitter as a product—is you get like these at mentions, these replies. You get the ability to just sort of jump into somebody’s network and grab their attention. And that becomes this thing that, you know, is a source of serious pain and trauma for a lot of people. Or this way of instilling a different kind of brigading behavior on the internet, that becomes very foundational to how we do this thing.

At the same time, it is very much the thing that myself and other people on the service found as like—this is why this thing is revolutionary. Like, I can jump into this other journalist’s mentions. Or I can jump into … there were, you know, the celebrities who are coming on the platform. Like Ashton Kutcher is this big superpower user, and like Joe Schmo, you know, sitting at work can “at” this celebrity. And boom, they just reply, and, like, you’re now texting with a movie star.

That’s the thing, to me, that I find so fascinating about the platform. It’s like: So much of what makes this thing so useful, so great, so able to drive culture, to actually have utility for folks in breaking-news situations or just, like, whatever big cultural moments, is exactly the thing that makes it so dangerous and instills these terrible behaviors.

Goldman: 100 percent. I think you’re completely right that this flattening of status across the graph is this novel feature of the product. Where it’s like, yeah, you know, someone who has you know, 500 likes on their tweet, feels as significant in the network as the biggest celebrity in the world. If, like that biggest celebrity, they have an equal chance of having a place in the firmament. It’s the beginnings of what we see dominating the media environment right now—which is this, you know, nicheification of all media. Where, like, if you’re famous to like 500 people, you’re famous. Like that you are a dominant, influencer news source for those 500 people, and that you are more meaningful to those 500 people than Ryan Gosling or whatever.

So that all started, I think, with Twitter. And the other point—like the point about why I think we were blind to the abuse and harassment stuff in those early days—is both because we adopted a bad content model from Blogger without realizing that the product had changed fundamentally. The types of abuse and harassment that could be implemented. But I think, two, one of the fundamental blind spots of technologists and people building services is that you’re up against such odds from so many people who don’t believe that the thing that you’re doing is worth a damn. And are just like, Why would anyone spend any of their time doing this? Like, it’s what we heard when we were doing Blogger. It’s what we heard when we’re doing Twitter. You know, it’s like—“This is just where you go to, like, tell people that you’re eating a taco. How is that worth anything?” And we’re like, “No, you don’t understand. This is what’s going to connect the human hive mind and, like, bring us to a higher level of consciousness.” That was like, you know, a true belief.

And when you’re that zealous about the mission that you’re on, which you almost need to be in a startup to survive, that zealotry blinds you completely to the downside risks that you’re producing. That is a dominant idea that you see—particularly in this era of the internet, like the 2000-to-2015 era of the internet—which is just, Any use of our product is intrinsically good. More of our product in the world is intrinsically good for humanity. We are good people. And therefore, we just need to get this product in the hands of more people. And if there are downside risks, if there are things that are happening that we don’t like, those are bugs, and we can fix those bugs. But that’s not the intended use. And so we shouldn’t judge the platform based on that.

Whereas what I now believe is that those downside risks—those bad things that happen—aren’t bugs. Those are just use cases that you enabled that you may not like. Like, all of those things are equally weighted use cases and things that your product enabled. And you need to grapple with that fact. Harassment of people in 2008. Gamergate. Like, those are all things that were enabled because of the system that you built and the choices that you made.

Warzel: Is there an original sin, so to speak? Or is it just a network of, or a body of, decisions that you’re making, sort of blind to these things? As you’re building the plane as you fly it? Or do you think that there is one thing where it’s like, If we could have gone back and nipped that thing in the bud…

Goldman: I think we dramatically understaffed the trust-and-safety team. And would routinely poach their engineers, because we needed them to keep the service running. You know, it was like, again, the service was not working. It was simply being crushed under its own weight. So, in a choice between, like, “Is the service going to stand up?” or “Are we going to have more people to build the internal tools that the trust-and-safety team needs to prevent abuse?” We chose the former 100 percent of the time. And that completely kneecaps that team’s ability to enforce rules or to make good policy. So that’s definitely one choice.

I do think kind of fast-forwarding to where we are right now, and you sort of already alluded to this, which is like: Why are these choices up to these companies 100 percent of the time? Like, why are these choices vested? Why is there no devolvement of authority from beyond the company’s walls? Now that doesn’t mean I want to create, like, a ministry of content moderation somewhere. But it does mean that there should be some checks on how the job is being done by these companies, by someone outside the companies themselves There should be some system, even if it’s just a scoreboard that says, “Hey; we’re seeing all the data in real time. We’re making assessments of where harm exists. Here’s our scoreboard of harm. Like, here’s like our doomsday clock. We think it’s like two minutes to midnight in terms of CSAM, or, you know, user abuse on your platform. And it’s moved from two minutes to, you know, 90 seconds in the last six months. Like, do with that what you will. But, you know, every month, we’re going to publish an update about how we think you’re doing.” And I think just simply having that sort of transparency would be helpful.

Warzel: Well, super candidly, it’s the thing that was so frustrating to me as a reporter. Because during this time—I don’t remember exactly when it was; it was around Gamergate—you’d see these high-profile cases of harassment. And it would trickle. Sometimes it would just be regular, normal women who were getting, you know, heaps and heaps of these, like, threats or, you know—

Goldman: Yeah. Death threats, rape threats. Yeah.

Warzel: Or things like that, that maybe bleed in the world. And what would happen is, they would report them to Twitter. And then it would be sort of like, “We’ve handled your claim, and we don’t see any problem.” You know, clearly there was no human in the loop. It was just happening. So it became this thing that myself and probably, I don’t know, 20, 30 other journalists ended up doing, where it was like we were acting as sort of a trusted safety team. Like an escalator person, basically, to flag it to, you know, PR and be like, “Hey, so, is Twitter going to enforce the rules here?”

And that’s the thing. Like, the media in this moment of, let’s say, 2015, 2016, 2017 became this, de facto layer in there, of like accountability. But there was, at the same time, I think it’s been retrofitted into this idea of like, “Oh, you’re trying to act as a censor, right? You’re trying to act as a thing.” And it’s like, “No, this is what’s very basic journalism stuff.” Which was like, you guys have a slate of rules that are changing all the time. And someone is coming and saying the rules have been violated. And, you know, as a reporter, you’re like, “Well, is the company going to enforce these rules? Or have the rules changed again?” Here we go. And then you’d see, no, “Such and such user gets banned, because reporter at X brought it up.” Or whatever.

And it sort of creates this condition that then, you know, evolves throughout the first Trump presidency. And you have, up even to now, which is like, “There’s this censorious media that’s just trying to get people banned from the platform.” This is a very load-bearing argument in the whole reason why Elon Musk decides to come in and want to purchase the platform: to, quote unquote “restore free speech” you know, to the people. But part of this frustration, I think, speaks to the idea of, like, not having any way to have accountability for these platforms. Enforcing their own rules is the thing. If Twitter had said, as they do now under Elon Musk, “We’re an open sewer. It doesn’t matter. We’re 4chan. Anything goes. There’s no rules.” Wild West.

Goldman: “It doesn’t matter. There’s no rules.” Yeah.

Warzel: There wouldn’t be reporters lining up saying, like, “Well, what’s going on here?” Because it’s like, yeah, no—I don’t want to be on this website that, there’s no safety net for it. And so I think that’s a really fundamental point that undergirds all of this. You exited the company in 2010, and I’m just curious…

Goldman: I was fired, but yeah.

Warzel: Fired. Okay. Why were you fired?

Goldman: Because I support, like … you know, Jack was fired in 2008. Which is something I supported as a board member. Jack comes back in 2010; it sort of precluded my ability to be employed there any longer. Not as, like, retribution, but just it was pretty clear that I wasn’t going to support Jack as a leader at the company.

Warzel: And this is an important part of all of this, too, at this time. Like, it is a relatively chaotic managerial; like, there’s just a lot of shuffling around. And like, the drama of Twitter. There are books written about it.

Goldman: Oh yeah; no. It was total chaos. Yeah.

Warzel: What did you think, in that time, was ahead for the company?

Goldman: The big debate was the sort of debate over the business model for Twitter. Of what was the team that we needed, and what was the path that we should pursue for a business model for Twitter. And Ev had a lot of—and me as well—we had a lot of hesitancy about the ads-only approach to the business model for Twitter. Because we had worked at Google. And you know, we’re not super wild about the user experience of content-targeted ads and advertising. And social media was like, you know, all this stuff about, you know, surveillance capitalism. And “you are the product.” Like, all of that stuff is true. It’s just true. You know, we haven’t come up with a better business model to replace it. It turns out that business model is both high margin and high volume. So like, you know, they get remains the dominant way to monetize content on the Internet, whether video text or anything else. But it just seemed kind of, you know, it seemed to be having a bad effect. And that was before we fast-forward 10 years and we see some of the negative consequences that that’s had for journalism, that’s had for, you know, for audience capture. All these other effects. So the fundamental tension was: How do we convert this company, in 2010, from a cultural success to a capital success?

And I think the fundamental mistake that we made in that time—I guess you go back to sort of the one mistake we made that I wish we could undo. The biggest mistake that we made was consistently comparing ourselves to Facebook on business terms. Like, when we were doing the run-up to the IPO, it was always like, you know. And even before the IPO, when I was there, like it was always like, “This is bigger than Facebook.” And that is just simply not true. Facebook has better data about its users, who are authenticated to be real users; they have more information; they do better surveillance across a suite of apps; they have a better ability to target them; they target on way better variables. Their ads product is just simply better than any that Twitter was gonna be able to build. And as a result, Twitter’s multiple was never actually gonna compare favorably. But its IPO story was that it was going to compare favorably.

So that was a big problem. That is what created the context for Elon to buy the company. Because, like, all of the early investors were able to get out of Twitter at the IPO time, or sometime in the intervening years, based on this idea that the multiple curve is going to be hit for Twitter’s business, and it’s going to exceed Facebook’s. The Street eventually realizes that’s not true; is under constant activist pressure from that time on to do something else. That puts a lot of pressure on Jack as the CEO, which he doesn’t enjoy. And that creates the condition for the company to be transacted to Elon. So if I had to go back to one mistake, it’s like it’s that whole run-up to the IPO and the way that the company is positioned as a business. That was the mistake.

Warzel: There’s this really important thing that happens in the 2010s after you leave, which is that this guy Donald Trump is somehow this really good power user of Twitter. He’s talking about Diet Coke, and people are like, “This is funny.” And he hosts this reality show that’s really popular on NBC.

And then he starts turning his eyes to politics. I always like to think that 2012 set the stage for politics happening on Twitter. These candidates are expected to be on Twitter. Lawmakers, when they get in, are supposed to bring people, you know, behind closed doors. Issue, you know, all sorts of their messages. The entire D.C. press corps gets so wrapped up, you know, secluded at times in this thing that, like, every micro-cycle drives editorial decisions in newsrooms, across everywhere. Donald Trump is like, “I can tap into that. I’m really good at this platform.”

Cut to, you know, November 2016. Donald Trump is elected president. This is also at a time when, as I’m covering it, the harassment stuff is really, really big. I was in contact with a lot of people who were working at the company at the time. And there was just this moment of like: I think we did this.

Goldman: I think we did this. Yeah.

Warzel: I think this is on us. What was your feeling?

Goldman: I agreed. I think Twitter was instrumental to Donald Trump’s election. I don’t think it is the only force. Or like, I’ve heard the 2016 election described, at various times, “like an airplane crash.” Like, you know, there’s multiple causes; you can’t really appoint it to just one thing. But, you know, the plane definitely crashed. It’s worth noting that on Election Day, November of 2016, I was in the Roosevelt Room of the White House when that happened. Because I was working at the White House by that time, and had been hired into the Obama administration because they had realized that they had lost the thread on using social media effectively after the 2014 midterms. I had this conversation with President Obama in the wake of the 2016 election, where he said … you know, all of the sort of, like, “Obama is very even.” He does not get very hot or very angry. And in the end, the day after the election, his role was going around to staff and bucking people up. To be like, “Hey, this is going to be okay. We’re going to be okay.” Which is pretty incredible.

And to me, he said, “You know, I’m not really happy with the results of this election.” I was like, “Yeah; me neither, sir. I think this is not good.” And he’s like, “And you know, there’s a lot of reasons for why that’s true. But one of them is because of you, because of Twitter.” And I was like, “No; I agree.” And he was doing that as a way to, like, start a broader conversation. About, like, “What do we need to do to think about the role of social media, and the role of the internet in our civic life and what it’s doing?” Which had been a topic that had interested him long before the election. But I agree with his assessment: that Twitter played an outsize role in that election.

What Trump realized before—you know, early, like eight years before we sort of really codified it into a thesis—was that the currency that mattered most in the contemporary environment was attention. That attention was the coin of the realm. And if you could command attention, regardless of if it was for good reasons or bad, you were winning. All you needed to do was to be able to command attention. And Twitter was very good for commanding attention, because you could say something outrageous, and that would get a lot of attention on Twitter.

And I think the way in which you need to understand things like meme coins and prediction markets are essentially derivatives on the attention economy. It is essentially ways in which people have figured out how to trade this fundamental commodity that runs the entire world now.

Warzel: So Trump’s president. And there’s this guy on the service, famous entrepreneur, Elon Musk. I think for me it was some period; it was sometime in 2017. It’s very clear in that time period of whatever, let’s just broadly say 2017 to 2020, that Musk himself, a power user of Twitter, sort of gets, and this is a very common thing that happens on the platform, is like…

Goldman: Gets captured. Yeah.

Warzel: People get captured by the platform. And they both like being good at it, affect it. And then it affects them, and the snake eats their own tail. Like, you know, it’s just an accelerationist situation, where you get to someone who is radicalized in their ways. Elon Musk then goes through this whole period of time in 2022 where he flirts with basically,, like, getting a board seat. Right?

He sort of reverses course and says, “You know what; actually, I’m going to buy this.”

Goldman: “I’m just gonna buy it.” Yeah.

Warzel: “And I’m gonna take it private, and I’m gonna totally rewire this thing. And I’m gonna bring this back to the glory days, to what the platform is supposed to be. Free-speech-maximalist ethos.”

Goldman: So I think a couple of things, for context. Elon is the best example for the “power of audience capture,” in the current environment. Because you’re talking about a person who is worth a trillion dollars, close to it. Can literally do anything. He’s got more power than most nation-states in the world. And yet, his engagement on Twitter is clearly one of the most important things to him in the world. Like, it is clearly very meaningful to him, and the feedback that he gets from his audience on Twitter is both so important to him that he not only goes to, you know, buy the product. You know, not only goes to buy the company that produces this product. But Joyce Carol Oates makes fun of him on Twitter for not liking things. The famous author Joyce Carol Oates, who’s an absolute—

Warzel: Yes, basically says he doesn’t like things that like humans like, right? Like, pets and reading and movies. Right.

Goldman: He doesn’t know. It’s like, why doesn’t he ever talk about his family? Why doesn’t he ever talk about movies? And like, just killer. It’s one of our great, just fire, Twitters, by the way. Just unreal that she’s great at the product. And he spends the next, like, two days or two weeks posting movies that he likes, or whatever. Like it clearly, it seemingly got to him that he felt the need to kind of respond to this, in this oblique way. Very curious effect of the product.

Warzel: So, Musk buys Twitter. Immediately, he walks in with a sink as a joke; “Let this sink in.”

He slashes the workforce. Creates a lot of actual chaos; fires a ton of people. And there’s this feeling for a long period of time that Twitter is actually going to die, right?

Goldman: Going to die. Yeah.

Warzel: And Musk is alienating advertisers at this exact moment. He’s, like, walking into meetings and saying stuff that makes them feel really weird and really bad about stuff. It seems that the service isn’t gonna hold up in those really early days. He’s also just, like, letting a lot of banned people back on the service. Donald Trump can return to Twitter after being banned post–January 6 and the platform doesn’t die.

Goldman: Doesn’t die. Still doesn’t die.

Warzel: In fact, lots of power users who are in my profession or in politics and stuff …

Goldman: Come back.

Warzel: Still using it, still enjoying it. Doing everything. It leads to the 2024 election. Elon Musk, really, while he’s doing all this, he’s in the tank for Donald Trump. He’s given out million-dollar checks to people. He creates, and a lot of people don’t remember this, an election hub inside Twitter or X. Which, he renames it X; forgot about that. He renames it X. There is this feeling he has put his hands on the scale. He’s used this platform that has his outsized political power. He’s sort of turned it into this political weapon that can be directed toward his particular ideology.

What does Elon Musk fundamentally understand about Twitter? What is the thing that he truly understands and gets, in ways that maybe no one else does?

Goldman: I think it is somewhat analogous to the Trump case—which is that attention is good. Because, you know, look—and this is always the throat-clearing part about talking about Elon that I hate doing. But, like, he did build a rocket company. He has achieved; he has built some real things. There’s no one who’s understood and leveraged as well the ability to turn attention into market capitalization better than Elon Musk. Like he really is, like, the, the best who’s ever done it. And some of the techniques that he’s applied there are personal mythology. You know, there’s a lot of personal mythologizing of Elon. He never sleeps. He sleeps on the factory floor. He’s a super genius who understands everything.

Twitter was a little bit risky for him, incidentally, on the “super genius who understands everything,” because it became very obvious in the early going that he didn’t know what he was talking about with regards to global web-scale design. Like, there’s a number of high-profile incidents where, like, people who actually understand how global web systems have been built. And Elon’s not built a website since like, you know, PayPal. Like we’ll go and be like, but you know, “What do you mean, like ‘There’s too many microservices’? Like, what does that actually mean to you?” And he clearly doesn’t have, like ,a practitioner’s understanding of that problem. But like, you know, the site does still run, right? Like, you know, it makes all these changes. The site does still run.

I think Elon has managed to parlay attention into market capitalization for his companies better than anyone else, because he’s figured out how to both skirt and in some cases just go over the line of what you can do.

I’ll put myself out there: I did not think his X experiment would work. I felt he would have to, like, kind of get rid of this. And one of the reasons was, I just didn’t think the numbers … I felt like the brand risk that he was going to take on. And this is before he went all-in for Trump, I was like: That’s just going to look bad for him, and people are going to start feeling bad about his cars, and they’re going to feel bad about his actual business. Like, why is he going to want that? Additionally, Twitter was so levered, in terms of the amount of debt that it had on its book, that it completely froze.

Like, it froze the banks who are holding the debt, because they couldn’t move it off their balance sheets during a time of rising interest rates. But he’s more powerful than the Street. Like, he is able to make the largest financial institutions in the world kind of dance to his tune, because his power is so unbounded by any normal constraint. There’s no one else who would be able to do that. Anyone else would get called. But no one wants to call Elon. I think it is a unique example of how to use attention for market power.

Warzel: So this is, to kind of put a bow on it. I am curious. Some of this has to do with Elon. But I think broadly speaking, I’m interested more outside of the realm of just him and his ownership and the logistics of it. But: Why won’t Twitter die? So many things happen to it.

Goldman: Yeah, man. It’s crazy.

Warzel: There’s so many outsize elements of negative influence. I mean, there’s positive influence, obviously. But when you look at it, this is a website that was, for a very long time, its power users referred to it as “the hellsite.” So why won’t it die? And I don’t mean from a funding standpoint. But like, why won’t it die culturally?

Goldman: Brother, wish I knew. I really don’t. I do find it interesting. There’s a couple theories. Once you own a piece of real estate on the internet, it does end up being pretty durable in a lot of cases. It does end up being pretty hard to dislodge people if they’re willing to fund the losses. Facebook—despite its misadventures in the metaverse—owns the friend graph still. Like, and despite all of, like, all of the youngs don’t use it, and like, you know it’s just for Boomers or whatever, it’s still like a core, load-bearing element of the internet, right? And for all of the, like, “You have to append ‘Reddit’ to the end of search queries in order to get the good results”—because you know SEO has gotten so jacked up, like Google still owns search and has owned it for 20 years, right? Like more than 20 years at this point.

And so it is hard to dislodge incumbents, because there’s no regulatory pressure. That is one of the things that traditionally has prevented incumbents from reaching a certain size. And market-consolidation forces are still one of the iron laws of the land. Like, the bigger just eats the smaller. Again, absent any regulatory pressure or paradigmatic reshuffling of the deck, à la the introduction of the consumer internet, writ large, like—you just don’t see broad-based dislocations in these mature markets.

So I think that’s probably the easiest explanation. I don’t think, like, there’s one magic trick that Twitter figured out that makes it so durable. I think that I was very skeptical of all of the folks who kind of came along, whether that’s Threads or Bluesky, and were just like, “We’re going to do this, but better.” I was like, nothing but better ever replaces the old thing. That’s just not how it works in mature markets. That can work when the market is evolving, but that doesn’t usually happen in a mature media market.

Warzel: What is the legacy of Twitter, for you? We’re talking about a lot of stuff that is really negative. There’s also a lot that’s like really fascinating. It is the most influential piece of technology that’s not a phone or something like that. Pieces like social media, in my life, to my career, to all these different things, right? I have this tortured relationship with it. I’m sure you do too. But in your mind, where does it net out for you? Is the legacy of Twitter positive? Is it negative? Is it indeterminate? How do you feel about it?

Goldman: I try not to be too solipsistic about it, because the stakes are higher than just my personal experience. There’s people. I’ve benefited tremendously from Twitter’s ascendance; financially as well as the things I’ve been able to do in my life professionally. But it is informed by the fact that, you know, I worked on Twitter early, and now Elon Musk runs it. And I worked at the White House, and my transition meeting was literally with Dan Scavino, who, like, tweets for Donald Trump to this day. And so in both of those—what is the legacy of that?

It is hard for me not to see it in the context of “What is the legacy of being an American in this moment in history?” Where you are raised in an era where you believe certain ideological things about the country, and certain things that are true, and a certain belief in its core institutions. And the idea of democracy, and the idea of a pluralistic system in which, you know, we’ve gotten past and are still contending with, but have gotten past a lot of the historical injustices that defined the first 250 years of this country.

And yet, we live in a moment where we are seeing, you know, just the most venal, worst-acting, worse aspects of American society in our lifetime. And the story of Twitter is embedded in that, very deeply.

And so it’s hard for me to reconcile both things at the same time. Which is that I still believe in the ideological ideas that underpinned the beginning of Twitter and also Blogger—which is that the internet should be used as a medium of self-expression, and that if we were to embrace it as such, people around the world would understand each other better. We would be able to experience more of one another’s lived world. And that would create more empathy and understanding, and that that is a good and virtuous project to be engaged with. And I’m proud of the work that I’ve done there.

And yet, what we actually have built—and what has actually been produced—is a platform that has created tremendous harm, and is in the current moment still being used to inflict tremendous harm. You have to wrestle with both of those things. And you know, you cannot just look at the things that you don’t like and say, “Those are unintended use cases that I never meant to happen.” Your intention doesn’t really play into the answer there.

Warzel: We’ve talked a lot about the cultural relevance of Twitter, and the fact that it can’t die. There’s this way in which it was very, very important for all these different genres of niche communities. You’ve got, like, Black Twitter was an absolutely foundational one for this type of culture, that really influenced not only the platform but culture at large in general. We’re talking about, you know, Elon and Trump and these guys, and being really good at using it. And I think that there’s this way in which, especially, two dudes who have so much history with the platform, overindexing its cultural relevance now. I think there’s a feeling that the culture outside of politics, and outside of edgelord, racist, Trumpian politics has actually moved on—when you have TikTok and these ascendant platforms where these communities are going, and where sort of like the bleeding edge of culture happens. Do you think that, I don’t know, that’s a threat to Twitter going forward?

Goldman: I think there’s an optimistic version of the story there. Which is that Twitter—by, in the Elon era, foregrounding so vociferously his parochial cultural interests, by being so overt about the cultural change that he wanted to see in the world—it creates its own natural backlash. Because it ends up becoming defined as a period in time.

And I think you’re right, which is that culture is being made in other places. Like when TikTok became ascendant; felt the same way about it that I do about Twitter. I did about Twitter. Which is just: I love TikTok. I think it is so exciting, that it is a place where culture gets made, and you see these little pockets of where it’s, you know, it feels like something that was made for you and your entertainment. I think it is a much more dangerous and weaponized version of social media than what we were doing with Twitter, because it is so precise and is so weaponized.

I do think Twitter could be sort of the thing where, when I got to the White House and was working in communications, I was surprised to learn about the centrality of like Morning Joe as a show that was really important for how people—the most powerful people in the world—thought about where content was being shaped. And I was like, “I’ve never heard of this show. What is this show?” People were shocked that I’ve never heard of it. But it is very important for how narrative is shaped. And Twitter could be something like that, where it is like, you know, a version of legacy media that is, in absolute terms, very, very small. And for certain cultural-making aspects been replaced by other platforms. But for some powerful part of the demographic—older male, you know, people in politics, people in news—it still is the Morning Joe of its time for the next 15 years.

Warzel: Jason, thank you for walking me through this.

Goldman: Yeah; this is great. Did we do it?

Warzel: We speed-ran it. But I’d really—I think that this is a pretty good “how a bill becomes a law that destroys civilization and the fabric of democracy and reality.” So I think we did do it. And I appreciate your insight. And yeah; thank you.

Goldman: Thanks so much.

[Music]

Warzel: Thank you again to my guest, Jason Goldman. If you liked what you saw here, new episodes of Galaxy Brain drop every Friday, and you could subscribe on The Atlantic’s YouTube channel, or on Apple or Spotify or wherever it is that you get your podcasts. And if you want to support this work, and the work of my fellow journalists at The Atlantic, you can subscribe to the publication at TheAtlantic.com/Listener. That’s TheAtlantic.com/Listener. Thanks so much for watching, and I’ll see you on the internet.

This episode of Galaxy Brain was produced by Renee Klahr and engineered by Dave Grein. Our theme is by Rob Smierciak. Claudine Ebeid is the executive producer of Atlantic audio, and Andrea Valdez is our managing editor.

Welcome to a Multidimensional Economic Disaster

2026-03-26T16:44:54-04:00

The global economy has become dependent on the AI industry. Trillions of dollars are being invested into the technology and the infrastructure it relies on; in the final months of 2025, functionally all economic growth in the United States came from AI investments. This would be risky even in ideal conditions. And we are very far from ideal conditions.

Much of the AI supply chain—chips, data centers, combustion turbines, and so on—relies on key materials that are produced in or transported through just a few places on Earth, with little overlap. In particular, the industry is highly dependent on the Middle East, which has been destabilized by the war in Iran. A global energy shock seems all but certain to come soon—the kind where even the best-case scenario is a disaster. The war could grind the AI build-out to a halt. This would be devastating for the tech firms that have issued historic amounts of debt to race against their highly leveraged competitors, and it would be devastating for the private lenders and banks that have been buying up that debt in the hope of ever bigger returns.

For the better part of the past year, Wall Street analysts and tech-industry observers have fretted publicly about an AI bubble. The fear is that too much money is coming in too fast and that generative-AI companies still have not offered anything close to a viable business model. If growth were to stall or the technology were to be seen as failing to deliver on its promises, the bubble might burst, triggering a chain reaction across the financial system. Everyone—big banks, private-equity firms, people who have no idea what’s mixed into their 401(k)—would be hit by the AI crash.

Until recently, that kind of crash felt hypothetical; today, it feels plausible and, to some, almost inevitable. “What’s unusual about this, unlike commercial real estate during the global financial crisis,” Paul Kedrosky, an investor and financial consultant, told us, “is all of these interlocking points of fragility.”

[Read: Here’s how the AI crash happens]

Perhaps the clearest examples are advanced memory and training chips, which are among the most important—and are by far the most expensive—components of training any AI model. Currently, most of them are produced by two companies in South Korea and one in Taiwan. These countries, in turn, get a large majority of their crude oil and much of their liquefied natural gas—which help fuel semiconductor manufacturing—from the Persian Gulf. The chip companies also require helium, sulfur, and bromine—three key inputs to silicon wafers—largely sourced from the region. In addition, Saudi Arabia, Qatar, the United Arab Emirates, and other regional petrostates have become key investors in the American AI firms that purchase most of those chips.

Because of the war in Iran, the Strait of Hormuz is functionally closed to most shipping vessels, stranding one-fifth of the world’s exports of natural gas, one-third of the world’s exports of crude oil, and significant quantities of the planet’s exportable fertilizer, helium, and sulfur. Meanwhile, Iran and Israel have begun bombing much of the fossil-fuel infrastructure in the region, which could take many years to replace. In only a month of war, the price of Brent crude—a global oil benchmark—has jumped by 40 percent and could more than double, liquefied-natural-gas prices are soaring in Europe and Asia, and helium spot prices have already doubled. The strait is “critical to basically every aspect of the global economy,” Sam Winter-Levy, a technology and national-security researcher at the Carnegie Endowment for International Peace, told us. “The AI supply chain is not insulated.”

The situation could quickly deteriorate from here. A helium crunch could trigger a shortage of AI chips or cause chip prices to rise. AI companies need ever more advanced chips to fill their data centers—at higher prices, the massive server farms, already hurting from elevated energy costs caused by the war, would have almost no hope of becoming profitable. Without these chips, new data centers would not be built or would sit empty. Astronomical tech valuations, and in turn the entire stock market, could collapse.

One industry’s precarious position isn’t usually everyone’s problem. Unfortunately, AI is different. The biggest data-center players, known as hyperscalers, are among the biggest corporations in the history of capitalism; they include Microsoft, Google, Meta, and Amazon. But even they will be pressed by collectively spending nearly $700 billion on AI in a single year. In order to get the money for these unprecedented projects, data-center providers are beginning to take on colossal amounts of debt. Some of this is done through creative deals with private-equity firms including Blackstone, BlackRock, and Blue Owl Capital—which themselves operate as sort of shadow banks that, since the most recent financial crisis, have arguably become as powerful and as influential as Bear Stearns and Lehman Brothers were prior to 2008. Endowments, pensions, insurance funds, and other major institutions all trust private equity to invest their money.

For a while, it seemed like every time Google or Microsoft announced more data-center investments, their stock prices rose. Now the opposite occurs: The hyperscalers are spending far more, but investors have started to notice that they are not generating anything near the revenue they need to. The data-center boom’s top players—Google, Meta, Microsoft, Amazon, Nvidia, and Oracle—have all lost 8 to 27 percent of their value since the start of the year, making them a huge drag on the overall stock market. And the $121 billion of debt that hyperscalers issued in 2025, four times more than what they averaged for years prior, is expected to grow dramatically.

All of the major players in this investment ecosystem are vulnerable. Private-equity firms are being squeezed on both ends by generative AI: During the coronavirus pandemic, they bought up software companies, which are now plummeting in value because AI is expected to eat their lunch. Meanwhile, private equity’s new investment strategy, data centers, is also falling apart because of AI. Blackstone, Blue Owl, and the like are sinking huge sums into data-center construction with the assumption that lease payments from tech companies will pay for their debt. In order to pay for their investments, private-equity companies raised money from major financial institutions—but now the viability of those lease payments is coming into question as the hyperscalers’ cash flow is strained. “There’s a reason to think we’re seeing some of the same 2008 dynamics now,” Brad Lipton, a former senior adviser at the Consumer Financial Protection Bureau and now the director of corporate power and financial regulation at the Roosevelt Institute, told us. “Everyone’s getting tied up together. Banks are lending money to private credit, which in turn lends it elsewhere. That amps up the risk.”

[Annie Lowrey: How to guess if your job will exist in five years]

The way the money moves is concerning, but so is the AI industry’s underlying business model. At every layer, the technology appears to decrease the value of its assets. The advanced AI chips that make up the majority of the cost of a data center? Their value rapidly decreases as they are superseded by the next generation of chips, meaning that the ultimate backstop for all of the data-center debt—selling the data center itself—is not actually a backstop. The way that AI companies make money when people use their products is also deflationary. OpenAI, Anthropic, and others charge users for using “tokens,” the components of words processed by their bots. This means that tokens are an industrial commodity akin to, say, crude oil or steel. But unlike other commodities, the cost of each token is rapidly decreasing owing to advancements in AI’s capabilities. Kedrosky called this “a death spiral to zero.” As the value of a token plummets, the value of what data centers can produce also falls.

The war in Iran affects data-center finances as well. Should energy prices continue to skyrocket, so will the cost of this already very expensive computing equipment, because it needs tremendous amounts of energy to manufacture and operate. And the war has exposed physical risks to these buildings. Janet Egan, a senior fellow at the Center for a New American Security, described data centers to us as “large, juicy targets.” It is impossible to hide these facilities, which can cover 1 million square feet. Earlier this month, Iran bombed Amazon data centers in the UAE and Bahrain. American hyperscalers had been planning to build far more data centers in the region, because the Trump administration and the AI industry have sought funding from Saudi Arabia, the UAE, Qatar, and Oman. Now there’s a two-way strain on those relationships. The physical security of the data centers is more precarious, and the conflict is damaging the economic health of the petrostates, thereby jeopardizing a major source of further investment in American AI firms. The Trump administration “staked a lot on the Gulf as their close AI partner, and now the war that they’ve launched poses a huge threat to the viability of the Gulf as that AI partner,” Winter-Levy said.

Plus, “what’s to prevent Iran or a proxy group, or another maligned actor, from tomorrow launching an armed drone against a data center in Northern Virginia?” Chip Usher, the senior director for intelligence at the Special Competitive Studies Project, a national-security and AI think tank, told us. “It could happen. Our defenses are not adequate.” State-sponsored cyberattacks of the variety Iran is known for could also knock a data center offline. You can build all manner of defenses—reinforced concrete, drone-interception systems—but doing so adds cost and time to already costly and slow construction.

Just a few things going a bit wrong could compound, all at once, into a cataclysm. To wit: Qatari and Saudi money dries up. Sustained high oil and natural-gas prices drive up the costs of manufacturing chips and running data centers. Already cash-strapped hyperscalers struggle to make lease payments on their data centers, while similarly strained private lenders suffer as all of the AI bonds become deadweight. Tech valuations fall, taking public markets with them; private-equity firms have to sell and torch their assets, putting intense stress on the institutional investors and banks. The rest of the economy, drained of investment because everything was poured into data centers for years, is already weak. Unemployment goes up, as do interest rates. “Bubbles pop. That’s the system,” Lipton said. “What isn’t supposed to happen is that it takes down the whole financial system. But the concern here is that AI investment isn’t confined and may spread to the whole economy.”

Even if Iran and the Strait of Hormuz don’t directly trigger an AI-driven financial crisis, the odds are decent that another vector could. (Remember tariffs?) Energy prices could stay elevated for years, because the targeted fossil-fuel facilities in the Persian Gulf will take a long time to restore. As the U.S. directs huge amounts of attention and military resources toward Iran, it’s easy to imagine China launching an invasion of Taiwan—a scenario that terrifies Silicon Valley, because it would halt the production of chips needed to train frontier models. That’s not even considering the single Dutch company that makes the high-tech lithography machines used to print virtually all AI chips, or the German company that makes the mirrors used in those machines. “There are too many ways for it to fail for it not to fail,” Kedrosky said of the AI industry’s web of risk. “All you can say for sure is this is a fragile and overdetermined system that must break, so it will.”

There are, of course, possibilities other than a full-blown, AI-driven financial crisis. Data-center spending could cool gradually enough that a crash is avoided. The revenues of Anthropic and OpenAI have been multiplying every year, which proponents argue means that generative-AI products are on track to eventually become profitable. But on the current trajectory, that would still take years, and there are good reasons to think that this growth will slow or halt. Notably, the main draw of AI tools is “efficiency”: Rather than growing their overall output and the opportunities available to people, executives are hoping that AI will allow them to make cuts to their business operations. The medium-term success of generative AI would likely involve millions of people being put out of work. The range of options seems to be somewhere from mildly bad to historically so.

Should the system break, much of the blame would lie squarely with the technology companies. The stakes of this build-out, from the beginning, have been framed in civilizational terms—a geopolitical race alongside an existential one. The winners will control the future and reap the rewards. At every step of the way, AI firms have appeared to prioritize speed above the physical security of data centers, supply-chain redundancy, energy efficiency and independence, political stability, even financial returns. And in that quest for unbridled growth, the AI industry has wrested ungodly amounts of capital from investors all looking for the next big thing, ensnaring the entire economy.

Simultaneously, these firms have courted and even bent the knee to a presidential administration that has encouraged their “let it rip” ethos, only to watch as that same administration has plunged the industry into this emerging polycrisis. The AI industry was not made for the turbulence its leaders have helped usher in. The situation has grown so ungainly and untenable that, if Silicon Valley is merely forced to slow down, the viability of all this spending will likely be called into question in ways that could be devastating for many. In finance, being early is the same as being wrong. AI firms want the world to think they’re right on time. The world may have other plans.

How AI Is Reshaping the Battlefield

2026-03-20T13:00:00-04:00

Subscribe here: Apple Podcasts | Spotify | YouTube

Just how are powerful AI models being used in warfare overseas? In this episode of Galaxy Brain, Charlie Warzel sits down with Will Knight, a senior writer at Wired, to discuss the rise of autonomous weapons. From the origins of Project Maven to the recent falling-out between Anthropic and the Department of Defense, they trace what’s happening as artificial intelligence moves from summarizing documents to informing decisions on the battlefield.

How do these weapons work? What are the safeguards? Who decides what values get baked into these models? As autonomous systems become harder to avoid, where exactly is the line between human judgment and machine decision making? Warzel and Knight help explain how the Pentagon and Silicon Valley are more entangled than ever and where warfare goes from here.

The following is a transcript of the episode:

Will Knight: The U.S. government will talk a lot about the importance of AI reflecting American values. But what are those values, exactly?

Charlie Warzel: Right. Do we get to decide those values together?

Knight: Who gets this like that? Is that just [Donald] Trump, or is it just the heads of these companies?

[Music]

Warzel: I’m Charlie Warzel, and this is Galaxy Brain, a show where today we are going to talk about autonomous weapons and the future of AI in warfare. It’s a very strange time to be covering artificial intelligence. There is an active conflict in Iran in which artificial intelligence is being used.

There’s also this huge fallout between the AI company Anthropic and the Department of Defense over concerns about the use of their technologies. And there’s this broader feeling right now over the subject of autonomous weapons that, frankly, these companies have this very powerful technology that they are then handing to the military. And that technology is being used in ways that maybe these companies don’t feel like they have control over, and that these companies are certainly worried about.

There are so many moral, legal, ethical concerns. There’s so much that we don’t know about how these models actually work, the decisions that they make. Whether they hallucinate, whether they can fail at rates different and more concerning than humans, whether humans are in the chain making these decisions, whether there are the appropriate safeguards, whether the ideologies of these companies and their leaders actually fit the ideologies of the military, and whether that conflict is something that we should all be having a real conversation about. It’s such a messy, scary moment, and Silicon Valley is totally caught up inside of it.

So I asked Will Knight to join me to talk about all of it. Will is a senior writer for Wired, and he covers artificial intelligence and he writes their AI Lab newsletter. Together, we get into the nuts and bolts of all of this: who the players are, what this technology can actually do and not do, and what the future of AI warfare might look like in a moment where we are constantly escalating for fears of being left behind. Will joins me now.

[Music]

Warzel: Will, welcome to Galaxy Brain.

Knight: Thanks for having me.

Warzel: So I want to start, and I want to ground this conversation in a little bit of history, right? Because there’s this long and somewhat amazing history of military innovation. The military building weird or niche things that then diffuse into a broader culture. Everything from radar to wristwatches to GPS to duct tape to the internet, right? So it is not uncommon, broadly speaking, for the government to fund ambitious, vague technologies for the battlefield that then have all these other applications. And it’s also not uncommon for them to partner with outside companies to do this work. I would love to go back to, if you wanna go even further than this, we can—but maybe starting at Project Maven and the announcement of that in 2017. The original mission is, you know, use computer-vision algorithms, right? To analyze drone footage, detect objects. And the quote was, turn data into a quote, “actionable intelligence and insight[s] at speed.” What’s the backstory of that?

Knight: Yeah, well, I mean, I think that the backstory goes much further back in some ways, right? The sort of architect of lot of the U.S. thinking on AI was Ash Carter. This was the sort of pre-ChatGPT era. But in the sort of deep-learning era, it seemed like, and you can see why, it would fit very well with military things—like targeting, like these image algorithms that could spot things. It seemed like that was going to be a paradigm shift, and a really, really big one. At the time it was incredibly controversial. And it’s amazing how much things have changed. But you had big protests at Google, which had won this contract.

I think, you know, things have shifted. I think in some ways, that makes a lot of sense to me. I think the idea that you’re not going to use something like AI in the world of defense seems kind of absurd, right? It’s like saying you’re not going to use software.

Warzel: I think you have to look at it too, right? Going back and looking into the Project Maven backlash initially, and the fear at places like Google from some of these engineers and folks at the time, was: That all sounds fine, potentially, but I’m worried we’re going to end up making autonomous weapons. Right?. We know, sort of, where the path could lead here. I think what’s really fascinating about those initial protests is it was over this idea of where this all leads. And it seems like they’re correct. Like, this is where it is leading to, right? It’s not just these vision algorithms.

Knight: Well, I guess I would say there’s always been a very long-standing principle about sort of how you sense and make decisions, and that being really critical in military conflict. You can see how the use of a vision algorithm would be very important in that. And so you can sort of see this trajectory toward more and more, you know, automating more of that process. But one thing I would say is—having many times interviewed people at the Pentagon, and people in the armed forces—there are very good reasons why people on the ground do not want to hand that off. And commanders don’t, as much as I think the public thinks. Those systems are unreliable. And the idea of handing off decisions about taking other people’s lives, you know, making terrible accidents, is not taken lightly.

And I think there is very good reason why it is not the case that that is going to be rushed in. For a long time we’ve had systems that are, by many people’s definitions, fully autonomous. So you have systems that will fully autonomously destroy a missile that’s trying to hit a ship. You know, there are these systems that will shoot those down, and they’ll have shot it down before the person can react.

We do have guided missiles that will go into an area. And so you would have a set area where you can configure the parameters where you know there’s only going to be enemy combatants. Those are extremely expensive: the rare, what they call exquisite systems. One of the things we’re seeing with Ukraine is, you know, off-the-shelf drones weaponized. And the way software can control those means that it becomes a lot easier for autonomy to be deployed.

One of the things that people know is going to be really sort of game-changing is swarms of them—like lots of them working together—because it’s much harder to counteract 20 of them coming in to try and attack your tank. And then you can’t have 20 people operating their own drones and defending it. So you’re going to have situations where autonomy becomes slightly more … it’s going to become harder to prevent it, I think, in some situations beyond those of purely defensive ones.

Warzel: I want to drill down a little bit in how these warfare models work. Specifically, to the extent that you can help us understand, because I think demystifying all of this is very important—down to the, you know, “explain it like I’m 5” version. But to the extent that, like: How do they work? How many humans are there in the chains of commands? You know, how are these things, the nuts and bolts of them, working? How they identify targets, safeguards, all of that.

Knight: Yeah, these are classified systems. So we don’t have—I don’t have—a hundred percent visibility on it. The way I understand it: It’s slightly less that they’re feeding in a ton of information and saying, What should we do? You would see a map, and you would maybe have assets on it, and you could ask the language model to ask questions about the, you know, maybe the signal’s intelligence that was related to that particular area. And I think you would have all these sort of different resources, where you could ask things about it. So I think it is a little more sort of, you know, being kept at arm’s length in terms of the model-making calls. I think that they’re not crazy, and they’re not stupid about, like, how they ought to maybe not rely on the system a high level. But if a language model were making some errors, and maybe the user didn’t check carefully, could that lead further along to erroneous decisions? So it raises questions about how people are trained to use those. How much they rely on this. What sort of trust it kind of maybe inspires in people they shouldn’t have.

Warzel: It does feel like there’s probably somewhat of an analogy, if I’m gathering this correctly, between the idea of … we’re journalists, right? We say we get this series of spreadsheets, or this list of data, and want to use a language model. We let it look at this corpus of information, and it provides insights. It’s not saying, This is your story; write this. It’s saying, no, What I identify here are some patterns that we have seen. That is more of what we’re talking about here when we’re talking about that intelligence.

Knight: Yeah. That’s my understanding. You could compare it to maybe the law, or something like that, where clerks are using language models to analyze a lot more case studies. Or maybe medicine.

Warzel: You might not be able to answer this question, given your ability to see the systems, or know this—but I feel like it’s worth asking because so many people are using these generative-AI tools now. When you hear something like the Pentagon or the Department of Defense is using Claude, right? And it puts this thing into your head of like, you know, A lot of people are using these tools. Because they’re a friendly way to automate busywork, right? And they talk to you in this friendly way. And it can feel a little bit dizzying to think about it, right? And so I’m curious from you: Do you have an idea of how different this [military] AI would be from the commercial versions of this generative AI?

Knight: We do know that [Anthropic] gave the DOD a specialized version called Claude Gov. This is a great question. I was wondering about this, because if you prompt Claude anything to do with weapons or, you know, anything that might seem like related to military conflict, it will say, I’m sorry, I can’t help you with that. They’ve given it a specialized version, which I think—they’ve not disclosed this—but I think would have fewer of those guardrails. It would have to.

Warzel: Right.

Knight: But it’s an interesting question, I think—like, what does it mean when you unalign it? And I’ve played with some unaligned models. And they can behave in sort of surprising ways. They’re able to do more things, but they sometimes will push back. Even though, you know, some of that depends how those sort of restrictions have been removed. But you kind of have to give a model some guardrails to make it coherent. So I would be very curious, like, what the behavior of that model is. Does it sometimes actually reject stuff? Like, that could have happened, especially in the early stages. And maybe that, in pure speculation, but that maybe that has led to some of these kind of concerns about “woke” models and companies, or something like that.

Warzel: Right. And all of them, there’s a lot of debate between people who are real believers in this technology and those who are more skeptical of it. About personalities, right? And whether these things actually have personalities, or whether there’s just kind of like an emergent set of traits or ways they talk. But I think that that is fundamentally extremely interesting when you get down to this level of things. Like, you know, Claude is known for being a little more—compared to all the other frontier models—it’s a little more artistic. It has its own things.

Knight: Right; yeah. Its own vibe.

Warzel: And that becomes very interesting when you think about the ways in which these models are being ported into this moment. Where, again, they’re maybe not making decisions, but they are talking to a person. Or rather, giving an input in a very specific stylized, you know, human-sounding way.

Knight: There are really interesting questions about that. I think the truth is, most models have quite similar alignment. And I think that most of those things are fairly universal. And I think the question to me is, yeah, like: How does that really sort of change when you put it in a military setting? If they’ve sort of buttoned it down very much and it’s just doing some summarizing text—like here, ask a bunch of questions—that’s one thing. But if it’s trying to sort of do this sort of parasocial, you know, stuff that chatbots do normally? That is kind of strange. And that does really affect the user experience, the user expectations. Of, you would probably have much higher trust in a system like that if it were pretending to be a person, maybe, when you shouldn’t have.

Warzel: We’ve got at this a little bit, but I think it’s just worth drilling down even more on it and break down what we mean by autonomous weapons, right? Because that is a terrifying-sounding phrase. And the point of demystifying all of this, you know—it’s not always clear cut, right? Do you think when we talk about autonomous weapons, it’s best for people to think of it almost just like they have the ability to make split-second decisions in anything? And that’s sort of the place where we should work from? Or do you think there’s actually more specificity there? That, like, autonomous weapons are actually, as well, the thing that we’re all worried about, right? Which are weapons that are making broader choices about target acquisition or engagement in some kind of way.

Knight: The DOD has been clear that they want to always, at the moment, keep somebody making that final decision to some degree, right? That’s the thing. So you can have things be autonomous, but you have a person make that decision. But they issued a directive in 2023 on autonomous-weapons systems. And that very clearly spelled out that there’s no restrictions on developing them.

I’ve looked at autonomy in places like self-driving cars, in robots, and in the AI industry. Like agents, right? I have not come across anywhere that is more careful about what they do than the DOD in spirit. And that doesn’t mean they’re gonna get everything right. But I think that, I wouldn’t want to … I think it’d be a mistake to sort of portray the people in the armed forces as wanting to do autonomy for the sake of it. And I think that if they could keep their finger on the trigger, they would want to do that 100 percent.

Warzel: I think that that’s very important context. And it’s also very important to the sense of “there are so many different people in this chain”—from the people at the very top who are giving, you know, press conferences and who may be more ideological, down to the people who are nuts and bolts, pulling triggers.

Knight: But you’re making a great point when you talk about the leaders of countries, including our own, that will bear a lot of responsibility for how quickly these things get kind of rushed in. I mean, this is one of the things with adoption of AI—that it’s seen especially in the U.S. as a way to kind of regain parity and overtake China. Because the U.S. by some dimensions, has the biggest military. But by some dimensions it’s at a disadvantage compared to China. So leading in AI, like—this is one of the real reasons why AI is such an important thing for the U.S. government. And I think there are a few questions. Like how aggressively you want to rush that in; how much you want to throw safeguards out the window.

Warzel: All of this comes down to: When we’re talking about this, it leads us to the DOD-versus-Anthropic situation that has played out over the past couple of weeks. Because I do think it speaks a little to all of this as well, right? Which is the understanding that this technology … the people making this technology, in some senses, have strong feelings about how it should be used and what they do not feel comfortable with the use being. And then you also have a current administration that feels very strongly that they, speaking of autonomy, must have full autonomy over how things should be deployed. Can you just walk through the particulars of that fight, for someone who’s only very, you know, basically paid attention to it?

Knight: Okay, yeah. Well, I mean, there are some elements we don’t have the full details of. Broadly, a few matter of weeks ago, the Pentagon wanted to change the contracts that it signed last year with all the big AI companies. What was in there previously, that Anthropic didn’t want to change, was specific prohibitions against using it for mass surveillance of U.S. citizens, and for autonomous weapons. So the Pentagon says, We don’t want to do that. So let us change that. In a way, Anthropic just decided that was a hill they’re going to die on. And I think a lot of people in the AI world are very conscious of safety and conscious of the sort of moral questions around it.

And so OpenAI stepped in; so they’re going to take this contract. But they want to have some safeguards. But the question is: What are those safeguards? And the question is: What situations is it acceptable to have something that might fail X out of a thousand times? Like, we don’t know how much that failure rate is. And what times isn’t it? And I think if it’s gonna make the difference between telling somebody, giving some of the information that they might make a lethal decision, I think this is the question the public should have. Like, where is that line being drawn? And especially as this gets deployed more widely, I think it’s been sort of quite tentatively implemented so far. And as that gets more capable, and these errors can sneak in in surprising ways, like where is that? Where should we, and where shouldn’t we? And yeah—that’s sort of the question that’s not been answered.

Warzel: You know, one thing that we didn’t say was that Secretary of Defense Pete Hegseth and the administration are designating Anthropic—as a result of some of this—as a supply-chain risk. Which a lot of people have said is an extreme overstep of the bounds. So there’s that element, which seems, to my sense, very irrational. And then there is the understanding of the actual nuts and bolts of the contract, right? Which is, Okay, maybe this doesn’t make sense for us.

Knight: Yeah; he had this moment where he realized that [with] the model, you had to get permission to do all these different sort of things. So it does sound like either the model was not as unaligned as they might’ve wanted, or the safeguards—the sort of checks you had to go through—the way the system was designed wasn’t working for them. But yeah; that reaction. Like, to put that in context, that’s only ever been applied to Chinese companies like Huawei, who are accused of operating for the Chinese government. And to take that action—not just on a U.S. company, but on one of the most important, by their own measures, U.S. companies that there is, working in this incredibly important technology—seems amazingly destructive to me. And it’s a very, very extreme action. But that’s sort of … yeah, I guess that’s just how they roll.

Warzel: Something that’s really important here, to my knowledge—and it speaks to this whole Anthropic dispute; it speaks to almost everything we’re talking about when it comes to the companies interfacing with the government. They create these powerful models with these capabilities that they understand will be used in war fighting. And they try to set their own boundaries to some degrees: guardrails, safeguards, whatever it is.

Then they have to hand it over, right? And it becomes a bit of a black box. There is this lack of understanding in specifics of how this stuff is being used on the ground, because some of these operations are extremely classified. Some of these uses are extremely classified.

As all this Anthropic fallout is happening in the news, the United States goes to war with Iran. There are bombs dropping. And so there’s this moment, a week or so ago. There’s these first reports coming out in the press that perhaps the United States was responsible for this deadly Tomahawk-missile strike on an Iranian elementary school. There was all this speculation on X, in the media. You know, speculation is what it is. That we’re hearing about these systems like Claude being used for targeting, potentially. And here is a potential targeting failure.

Now again, making no link or connection—this is all just pure speculation—but it speaks to this idea of, you know: Would a company like Anthropic even know if their technology was being used? Or any company; let’s not even use Anthropic. Would any AI company know in a situation similar to that—let’s not use that one—if Were we involved? And I think that that’s what I mean by “the black box.” Which is, you know, you kind of have to give it over. And I doubt that they’re getting that visibility, right?

Knight: Yeah. You know, like the company that made the missile that they fired didn’t know anything about it. Didn’t have any say after that was used. I think, as far as I understand, it is not the case that Claude would have suggested that as a target. That’s my understanding. I think that’s right. I don’t know how much they rely on Anthropic or these other companies to help build out new applications, new uses of those models. Maybe partly Anthropic wants to have more visibility on that, because they believe they know how to do it more reliably. That would be my suspicion. And I think it would sort of behoove the DOD to be aware of that, as well. These people understand how these systems can fail. So you’d want to try and work with them. Remember: Right now, what we’re seeing in the AI world that’s taking off like crazy is a next generation of these tools. And these are agentic AI.

Warzel: Right.

Knight: And I know that’s buzzword; they genuinely are. You’ll ask a model, or ask one these tools like OpenClaw, to do something. It will write code, and it will go through a bunch of steps. And that does raise more risk of unintended consequences, because those systems are doing things, and because they’re going through a bunch of steps. I use OpenClaw. It’s amazing, and it is really capable. You can see the allure and why it’s the direction things are going to go. But it will do things that I didn’t really want it to do, to try and get something done. So how you build those is going to be an incredibly important question, I think.

Warzel: I was preparing to talk to you, and I was having a conversation about this black-box potential element, right? And I said to this person, You know, they just have to hand it over, and that’s got to be terrifying. And the person I was speaking with pushed back and said, Well, is it terrifying? And their rationale was past defense contractors: They’re probably not terrified to hand over the missile or the hardware or whatever it is—or the database-infrastructure system—to the United States military. They’re probably like, they know where their purpose is in this chain.

I was thinking about this and how the fundamental difference, to me, feels exactly what you’re saying. It’s a fundamental difference of artificial intelligence. Because a company like Anthropic, in this situation, was deeply concerned in this way that others may not have been in the past. Because it’s not a technology that does one discrete thing. There’s just an unpredictability here, becoming more and more unpredictable with each iteration of this type of technology. And the intelligence [it] gives. Its whole purpose is to give a world of options without a complete and specific control. And what do you make of that idea?

Knight: I think what we’re likely to see in this and other critical situations is the development of new kind of approaches to engineering, actually. And how reliable those will be will be an interesting thing to watch. So a good analogy is self-driving cars. Those systems also use nondeterministic technology. So, like, a vision system that you can’t predict actually the decision it’s going to make, or the output. What it classifies something as. And it has to do all these things on the fly, where, you know, it’s potentially life-or-death situations. So those companies have had to design conventional engineering around that to make sure there are safeguards in all these different steps. And that’s taken years. And those cars can still only go … I mean it’s amazing, and they’ve made great progress only going in limited situations. So, you know, I think it’s doable. You can do the sort of safety-critical engineering with these kinds of more nondeterministic, more unpredictable technologies. The challenge is when that technology is advancing so quickly. Like, how do you wrestle with that? Like, it’s going so fast; it’s hard to implement those. One of the other differences with Anthropic is it’s just not a missile-making company. And people signed up because they wanted to make the world safer with AI, right?

Warzel: You’ve reported on concerns from people in the industry. But also the outsiders, the watchdogs. The people who are concerned about AI safety, but also just the safety, in general, of autonomous weapons. Can you run me through a list of some of the practical concerns that they have?

Knight: There are some who just believe, from a moral standpoint, that handing off … We should draw a line and say, “We can’t hand off.” So like: use of chemical weapons or something. They’re just worried that there’ll be this sort of huge expansion of autonomous systems, and that’s just morally wrong. And that’s a position that I think is a reasonable one to take, and can be debated.

The more practical issues that people would have are absolutely, like, just the incredible unreliability of systems. Especially, like, if you think of a self-driving car that maybe works on the streets of San Francisco, where it’s been trained for years and years right now. But if you’re taking it to unfamiliar battlefields, the probability of mistakes go up. The idea that these systems would misidentify a noncombatant and take their life is a huge concern. And I think there are people on the very technical side who would just, I think, worry about the—as you were saying—the reliability of systems that incorporate these more inherently unpredictable technologies. They might seem like, Okay, this only makes a mistake like once every 1000 times—but that they can kind of compound and cause problems in unpredictable, unpredictable ways.

Warzel: We’re talking a bit here about the idea of fail rates. About the idea of these models pulling in all this different information and making these insights that are either hallucinations or having some kind of failure inside them. But humans, too, very clearly fail all the time. And so what is it about the AI of it all, and these models, that makes this a scarier proposition to certain people?

Knight: Right. Yeah, I think what we should be worried about—so you’re right; people make mistakes. And the way the military is set up is to kind of cope with that in many ways. To try and have this chain of command, and so on, that minimizes the risk of that. But the thing that we should be concerned about is that these models, one, pretend to be very human, and they really seem human. And then they will just fail in totally spectacular, unexpected ways. So it’s like you’re talking to an infantryman, a soldier, who gives you the right answer again and again, and you’re like, Guy’s so good. And then all of a sudden he’s like: completely batshit-crazy answer. And so that kind of unexpected thing is, I think, what is most worrying. And I think that’s what the AI companies know. One, it can fail in sort of really unexpected ways sometimes. And then that people come to get really fooled by how human it seems.

Warzel: How much of this do you think is inextricable to the concern, broadly, to what has developed as a pretty reactionary culture recently in Silicon Valley?

You have people like Palantir CEO Alex Karp. He was on CNBC last week, and he said, quote, “This technology disrupts humanities-trained, largely Democratic voters and makes their economic power less. And increases the economic power of vocationally trained, working-class, often male voters. These disruptions are going to disrupt every aspect of our society.” Now, that’s not about autonomous weapons or AI technology specifically, but it is about instilling or having an ideological value inside of the company, and any broader mission statement. Do you think that that is part of the reason why people are having such a strong reaction? Does that change the valence of these conversations and the fears?

Knight: I don’t think it has, as yet. We started to see people sign open letters, and quite prominent people at some of the other AI companies backing up Anthropic in public. We haven’t seen that much of a discussion of where the line should be drawn, whether it’s in the political world or that much sort of public debate or discussion about it. And it’s going to be—you’ve got just a few people who are making really, really important decisions. Whether it’s the DOD or it’s the CEOs. So I would not be surprised if we saw much more pushback on it in many ways, whether it’s taking jobs, whether it’s being used in the military. I think we don’t have a great picture of what’s going on in employment, in the labor market, whatever the CEO of Palantir might say. But I think if it does start to do that, people will start to ask why, you know, these companies—that have honestly built tools by slurping up copyrighted material—why they own that. Why they are sort of in charge of disrupting everybody’s jobs. And you know, I don’t know that that’s the right position, but I think we could see much more of that.

Warzel: It seems inevitable to me, too. Because, as we said, this isn’t an Excel spreadsheet, right? It’s dynamic in this sense. It has to be instilled, whether it’s Anthropic talking about Claude having a constitution, right? Some of these companies talk about it as having values. Elon Musk has XAI. He talks about it having anti-woke values. And so then when you port this over into, yes, these systems are going to be in the chain. A chain that has decisions that ultimately can lead to people being killed or geopolitically significant events of war. They are not doing that thing of handing over a missile that is inert and saying, Use this as you will. These are dynamic systems.

Knight: Right. Yeah; I think that the constitution, personality, morality of those models is a whole other thing that this does raise. And it does. It is really interesting, right? You know, like the U.S. government will talk a lot about the importance of AI reflecting American values. But what are those values, exactly?

Warzel: Right. Do we get to decide those values together? Yeah.

Knight: Who gets this like that? Is that just Trump, or is it just the heads of these companies? We’re on a trajectory where the beginning of use of AI and it’s not going to be just summarizing things very plainly, probably for a long time. And so, yeah; whose values does that, who do they reflect? What are you trying to put into those? I mean, I think it’s somewhat encouraging that if you look at any model—you take a Chinese model or a U.S. model—most of what they’ll say you should and shouldn’t do are quite similar.

Warzel: There’s a great Bloomberg story by the reporter Katrina Manson, who notes in this piece about Project Maven, about a lot of this AI-warfare technology. She writes, quote, “There’s a palpable sense within the Pentagon that things aren’t moving fast enough. Despite the show of force in Iran, officials worry that the U.S. is at risk of falling behind. And officials are already looking past the Middle East to potentially bigger conflict. As one person familiar with the U.S. operations puts it, ‘Iran is an amazing precursor to what could happen with China over Taiwan.’” Now, that’s not your reporting, so I’m not going to ask you to talk about the veracity of those types of claims. But in your mind, what comes next here? This is obviously the beginning of something.

Knight: Yeah, well. So just to say that that is the tenor of a lot of conversations around the Pentagon and Washington. On the one hand, you can say, Well, that must reflect that this is where we’re heading. And there’s a lot of scholars who believe that, you know, some kind of conflict is inevitable. There’s also a sense in which those things can become self-fulfilling, and arming yourself to the teeth with technology historically has contributed to conflict. Like the First World War, for example. And so what comes next, I think, is sort of up to us and China, right? Just assuming that’s gonna happen, and sort of trying to arm yourself and prepare for that, might cause you to think that’s the only solution, which I think would be a devastating and terrible one. I would hope that there’s a different path that is going to not lead to that.

Warzel: Will, thank you so much for coming on the podcast, helped demystifying some of this stuff. It’s important work. Thank you for your reporting and all of it.

Knight: Yeah. Thanks for having me. It was very fun.

[Music]

Warzel: That is it for us here. Thank you again to my guest, Will Knight. If you liked what you saw here, new episodes of Galaxy Brain drop every Friday. You can subscribe on The Atlantic’s YouTube channel or on Apple or Spotify or wherever it is that you get your podcasts. And if you want to support this work and the work of my fellow coworkers, you can subscribe to the publication at TheAtlantic.com/Listener. That’s TheAtlantic.com/Listener. Thanks so much, and I’ll see you on the internet.

Maybe Turning War Into a Casino Was a Bad Idea?

2026-03-18T17:05:46-04:00

On March 10, the journalist Emanuel Fabian reported on a missile that had been launched from Iran. The warhead hit an open area outside Jerusalem, which Fabian confirmed by speaking with rescue services and reviewing footage of the explosion. He wrote a short post on The Times of Israel’s live blog and moved on.

Meanwhile, gamblers had wagered millions on the unfolding events of the conflict. Fabian’s post became the subject of a major dispute on Polymarket, a popular prediction market where people can bet on the outcome of almost anything. The site had allowed users to guess when Iran would initiate “a drone, missile, or air strike on Israel’s soil”: More than $14 million was riding on whether such an attack had happened March 10.

[Read: America is slow-walking into a Polymarket disaster]

People started reaching out asking Fabian to change his article. Some argued that Israel Defense Forces had not officially mentioned such an attack occurring on that day, and others said that the explosion he had reported was the result of a missile being intercepted, which according to Polymarket’s terms wouldn’t count as a strike “on Israel’s soil.” Confident in his reporting, Fabian did not amend the text.

And then he began receiving threats. “You will discover enemies who will be willing to pay anything to make your life miserable—within the framework of the law,” one person wrote to Fabian before adding, “As far as I know, there are also some people who don’t really care about the law, and you’re going to make them lose about 50 times what you’ll ever make.” Much as athletes have faced threats and harassment from fans with money riding on a game, prediction markets are now creating incentives for gamblers to target all manner of people with inside information or some influence over major events. Polymarket did not respond to my request for comment, but wrote on X: “This behavior violates our Terms of Service & has no place on our platform. We’ve banned the accounts for all involved & will pass their info to the relevant authorities.”

[Read: A technology for a low-trust society]

Prediction markets like Polymarket post online using the language of news wires and position themselves as a new and unbiased source of information, yet this story suggests that these sites are having the opposite effect: They make it harder for news gatherers to report the truth. Yesterday, Fabian spoke with me from southern Israel about what it’s like to be in the center of this controversy while simultaneously trying to cover a war. What he described was yet another way that online events are twisting the very nature of reality—leading Fabian, for just a split second, to doubt what he had seen and heard.

This conversation has been edited for length and clarity.

Charlie Warzel: How are you doing?

Emanuel Fabian: It’s been an overwhelming few days. I’ve been busy reporting on the war, and on top of that, I’ve been having to deal with the police and my family and all of these death threats and harassment. So it’s been a lot.

Warzel: Are you still getting death threats now?

Fabian: I’m not. They stopped almost as soon as I went to the police. Since the article I wrote about them went up, I haven’t received anything.

Warzel: You published your original blog on March 10. People began reaching out after that. But when did you make the connection to Polymarket?

Fabian: It took me a little while. When I got the first email about the missile impact, I thought the question [whether the missile exploded was intercepted, scattering shrapnel] was so odd, because it was such a minor, inconsequential detail in the context of a big war. The next day, I got a second email with the exact same questions and thought it was very strange. My theory was that it was either Iranian bots or agents trying to get information out of me. I did entertain the idea it was related to gambling, but I didn’t find the bet initially when I searched online.

The way it clicked for me was that I started to get replies on X and WhatsApp with similar questions like, Hey, why haven’t you updated your story? I figured something was up. I looked at the X profiles and could see they were very clearly Polymarket gamblers. At that point it clicked, and soon after I found the actual page itself for the March 10 bet on whether Iran would strike Israel. It was stuck on March 10 and the market hadn’t “resolved,” or paid out. All the comments were people going back and forth, many linking to my little story and other articles. Overall, I got at least 20 different messages across email, X, WhatsApp, and Discord.

Warzel: You said a contact from another media outlet also reached out to you at this time and suggested they had gotten a tip that your story was wrong. Was this person involved in the gambling as well?

Fabian: They messaged and said, Somebody I know told me there’s a mistake in your story; could you correct it? He thought he was doing both of us a little favor. I told him his acquaintance was likely betting on this on Polymarket. My contact went back to him, and he confirmed that not only was he betting on it, but he offered to give the person money if they managed to persuade me to change my story. It’s all insane. Obviously, the colleague told him off. But I’m losing my mind at this point. This is like the most tiny, inconsequential detail in a small news item.

Warzel: So you decide to call these people out on X. Did the harassment pick up after that?

Fabian: It did. A lot. I thought calling them out would shut them up and get them off my back. I wanted to be proactive because I realized, if I give into these people, it shows I can be manipulated. This will be just the beginning, and they won’t stop trying to bully me in later stories. And that’s when it escalated—death threats, messages coming in at all hours of the night. Messages talking about my family, giving me ultimatums on how much time I had to correct the story. That’s when I went to the police.

[Read: Insider trading is going to get people killed]

Warzel: Did you ever think about changing the story?

Fabian: For a split second I did. I thought maybe I could be wrong.

Warzel: Like, doubting your reporting? After all, you’re making those calls based on other witnesses and videos online.

Fabian: I went and checked again with the military. It was a short item, but I reviewed footage of a large explosion. I had eyewitness accounts—people in the area who saw this massive explosion. And then I thought to myself, Why am I doing this? Triple-checking this minor incident, bothering the military again over an explosion in the woods? I did the reporting, and this was the judgment call I made. I think it was accurate, and I will leave it at that. I don’t need to doubt myself about what I published, especially because this is not something that anyone normally would care about unless they had a financial stake in the outcome. As an event in this war, it is not particularly newsworthy. This missile exploded in an open area. It’s 150 words in the live blog.

Warzel: Do you think this fiasco will stick in the back of your mind as you continue to report on the war?

Fabian: Yes. I think it already has. Since then, whenever I report on something, I feel it in the back of my head: What if the Polymarket bettors are betting on this tweet? Or on whether I’m giving an interview about Polymarket? I’m not obsessing over it. Hopefully I won’t get threatened again. But the thought is there. What if they suddenly see this interview? Because I don’t know the way they’ve resolved the Polymarket bet yet.

Warzel: Wait, really?

Fabian: Yes, I’m looking now and the market is still not resolved. [The market “Iran strikes Israel on March 10 ?” resolved to “Yes” after Fabian and I spoke.]

Warzel: Did the fact that Polymarket kept allowing people to bet while this harassment was going on make things worse for you?

Fabian: It seems that a lot of people came into the bet as a result of my calling it out on X. When I posted about it, the market had $12 million in it. When I published my story on Monday, it had $14 million in it. Now it looks like it has $22 million. People are still betting and hoping it goes their way.

Warzel: Having been through this ordeal, what are your feelings about prediction markets in general?

Fabian: It’s really worrying. I think the gambling is a degenerate thing. The fact that people are betting on wars and conflict and people dying is gross. This is war, not a game. I think the more worrying thing is that we’ve seen harassment by bettors against athletes in sports for failing to perform. It seems now that we are entering a new age. I think there is a big risk of journalists using insider information to place a correct bet and win. I can tell you as a military correspondent that I’m exposed to confidential information that we can’t report. Now there are ways to exploit that. It wouldn’t surprise me if others have.

Warzel: Insider trading, one could argue, effectively makes prediction markets more accurate. Do you think these companies hope journalists and others will bet using privileged information?

Fabian: I don’t think they really want to combat insider trading. What I’ve heard is that those who bet on Polymarket either know the right answer or are wasting their money. [In a statement to The Times of Israel, Polymarket said, “Prediction markets depend on the integrity of independent reporting. Attempts to pressure journalists to alter their reporting undermine that integrity and undermine the markets themselves.”]

Warzel: Do you have advice for other journalists who may experience this type of betting-market harassment in the future?

Fabian: Go public. Don’t let the threats force you to change anything. Be honest. I think that’s the best way. It’s a bit stupid of these people to publicly intimidate somebody who can go and instantly tell 100,000 people what these gamblers are doing. That’s my advice. Because if you were to accept money or change your reporting, who knows how these people might extort you later on. If you change your reporting, it’ll be a mess forever.

Warzel: If you could sit down with the CEO of Polymarket, what would you tell him?

Fabian: I don’t know. I’d be honest and say I disagree with the notion of gambling on anything and everything. But if you are to keep these markets, they have to have admins who can decide on outcomes of bets or issue some kind of ruling. I think there just needs to be a lot more oversight and somebody actually vetting who these big bettors are to avoid insider trading but also to make sure this harassment doesn’t happen. But I’m not an expert on this. I’m more of an expert on where missiles land.

Doomscrolling Is Over

2026-03-14T06:47:00-04:00

From the comfort of my desk, I can see it all. A series of webcam feeds show me the sun setting over Tel Aviv and southern Lebanon. A map of the world, flecked with red dots, indicates that most of Europe and the Middle East are on “high alert.” I toggle a button on the map’s control panel, and the globe is instantly latticed with the locations of undersea fiber-optic cables. Below the map, a live feed of Bloomberg TV is running with the chyron Oil Extends Rout on Stockpile Talks. I scroll down and am greeted by walls of headlines, grouped into categories such as “World News” and “Intel Feed.” A “country instability” meter clocks Iran at 100 percent, while a different widget informs me that the world’s “strategic risk overview” remains “stable” at 50, whatever that means.

I am looking at World Monitor, a website that turns any browser into a makeshift situation room, and I love it. Built to look like a cross between a Bloomberg terminal and a big screen at U.S. Strategic Command, the site aims to display as much information about world events as possible in an assortment of real-time feeds. This is information overload presented as intelligence.

World Monitor was built over a single weekend in January by Elie Habib, an engineer based in the United Arab Emirates whose day job is as CEO of Anghami, one of the Middle East’s largest music-streaming services. “I wanted to extract the signal from the noise,” he told me recently. But what he really built, by his own admission, is a noise machine. Right now, the site pulls in more than 100 different streams of data, including stock prices, prediction markets, satellite movements, weather alerts, major-airport flight data, fire outbreaks, and the operational status of cloud services such as Cloudflare and AWS. The information is all real, but what exactly a person ought to do with it is unclear.

When Habib posted about the project on X, he was shocked by the response. At one point, tens of thousands of people were using the site at the same time; more than 2 million people accessed it in the first 20 days. Habib’s inbox filled with requests for new features as well as messages from venture capitalists looking to spin up World Monitor into a full-time business. Via GitHub, where Habib has made the code for World Monitor open-source and accessible to all, developers have made thousands of customized tweaks to the site and have translated it into more than 20 languages.

Obviously, people want immediate information on the conflict in Iran and the geopolitical and economic fallout from the war. But the site’s popularity stems from something else too. For the past year or so, extremely online weirdos—news junkies, day traders, social-media addicts, amateur investigators, guys who put up long posts on X about hacking their productivity—have embraced a meme about “monitoring the situation.” The phrase originates from a 2025 viral X post showing a jacked, arms-crossed, headset-wearing Jeff Bezos watching a Blue Origin launch: “The masculine urge to monitor the situation,” the caption says.

Like most memes, the bulk of situation-monitoring posts are ironic. They poke fun at the self-importance of the phenomenon. (“He’s not unemployed, he’s monitoring the situation,” one representative example reads.) Most of the people who make these posts are offering an enjoyable, winking blend of two perspectives: This is loser behavior and Dudes rock. Suffice it to say, World Monitor has thrilled this cohort, causing its fans to post things such as “BREAKING: you can now turn your laptop into a CIA command center.”

But this year, the monitoring jokes have taken on a different valence. The fog of the Trump administration’s wars has created an information vacuum that can immediately be filled on social media. Some of the people populating the world’s feeds are doing valuable work—the journalists and open-source-intelligence gatherers trying to confirm events and produce original reporting, for example. But they are outnumbered by propagandists, trolls, anxious commentators, war-market gamblers, and clout chasers who, apparently, became experts on the Strait of Hormuz overnight. These people post things such as “Hey babe, wake up, they just dropped a new war monitor.” They aren’t just monitoring the situation; they’re posting constantly about monitoring the situation.

[Read: This is what it looks like when nothing matters]

People treating war like entertainment seems like a logical extension of X, which has lost some of its real-time-news utility since Elon Musk took over and alienated many of the people who used to post there, and encouraged an army of edgelord users who treat the site like a 4chan board. (And people used to complain about the ludicrous ways that cable-news hosts vamped to fill 24 hours of coverage.) The meme speaks to something much bigger than that, though: Ours is a culture that has developed an insatiable need for instant information on all things at all times. Of course, we all live in saturated information environments, powered by constant connectivity and on-demand-answer services—Google, Wikipedia, chatbots. But I’ve also come to see all of this as a defense mechanism in an era of real chaos, when overlapping crises and technologies make the world feel unknowable and hyperreal.

The abiding feeling of 2026 is that too many consequential things are happening too fast for most people to follow, let alone understand. The United States invaded Venezuela in the night and captured its leader, Nicolás Maduro, 69 days ago. Renee Good was killed by an ICE agent 66 days ago; Alex Pretti was tackled to the ground in Minneapolis and killed by agents of the state 49 days ago. The last tranche of the Epstein files—millions of pages documenting Jeffrey Epstein’s dizzying connections to many of the most famous and powerful people in the world—came out 43 days ago. It’s been 22 days since the Supreme Court struck down Donald Trump’s tariffs. On February 4, a pseudonymous account believed to belong to an OpenAI employee snarkily commented that “Anthropic has the same level of name recognition among superbowl viewers as literally fictional companies.” Now the company is embroiled in a massive fight with the Pentagon; its CEO is on the cover of a forthcoming issue of Time. Yet most of these events have been pushed aside to make space for a war in Iran that the administration has hardly attempted to justify.

This is partly a consequence of our information ecosystem, which continues to evolve; more information is being created on more feeds, and through new products such as chatbots. Also, Trump’s reckless and erratic presidency has made reality move at online speeds. In the words of my colleague David A. Graham, the administration “can’t say why the United States went to war with Iran, and it can’t say what the goal of the war is. Now it can’t even decide whether the war is still going on.” The absurdity, the lack of pretense, and the senselessness all feel appropriate to the current age; as the writer John Ganz recently wrote, the war with Iran is “the first war that feels like it’s been launched by A.I: It’s all been done on a level less than thought.”

Monitoring is a reasonable response to all of this: It seems to offer a sense of agency. “They feel in control,” Habib told me when I asked why he thinks people like World Monitor. “They see everything happening in front of them, and it’s like, you know, watching a Bruce Willis movie.”

Yet this response to information overload is warping in its own way: People demand new news and commentary every time they refresh a feed. Taking even a short break can be disorienting when you attempt to rejoin a discourse that feels ever more self-referential and intense. Arguably, the best example of this dynamic is the Trump administration itself: Earlier this week, the official White House account on X published a video superimposing footage of the military bombing targets in Iran onto the 2006 Nintendo game Wii Sports. The account publishes stuff like this all of the time—and that’s exactly the point. The content outrages some people and delights others; publishing more of it advances the meta discourse that’s been layered on top of the actual news, drawing attention from the unfolding conflict itself. Because in reality, your attention can catch on only so much.

[Read: Believe your eyes]

This kind of thing is happening everywhere, constantly. If you’re not on World Monitor, you may be in a social feed, or in multiple social feeds, or trying to figure out which articles to tap into on a cluttered front page, or which newsletters to open in your inbox, or which podcasts to listen to at 1.3-times speed so that you can get to the good parts. The effect is not necessarily that you feel more informed; if you’re anything like me, you probably feel alienated, if not worse. Those who have chosen to try to keep up with the news cycle in 2026 are awareing themselves to death, as the writer Geoff George put it.

The situation brings to mind yet another grotesque online phenomenon: “gooning.” For the blessedly unaware, gooning is when maladjusted young men consume immense, overstimulating amounts of pornography and masturbate for hours on end to reach some kind of transcendent release. The comparison may sound absurd, but, as Daniel Kolitz wrote in a recent Harper’s article about the subculture, it mirrors the hyper-online monitoring behavior that I’ve been describing:

What are these gooners actually doing? Wasting hours each day consuming short-form video content. Chasing intensities of sensation across platforms. Parasocially fixating on microcelebrities who want their money. Broadcasting their love for those microcelebrities in public forums. Conducting bizarre self-experiments because someone on the internet told them to. In general, abjuring connective, other-directed pleasures for the comfort of staring at screens alone. Does any of this sound familiar?

The internet now implores us to binge as a default behavior: to watch whole seasons of TV at a time, to watch every football game simultaneously in quad-box fashion. We’re prompted to keep talking to the chatbot for answers or companionship; to let the AI agent accomplish task after task until we have built a website in an hour; to obsess in relentless, completist fandoms or go down rabbit holes. Total bombardment is partly a surrender to the internet and its logic and algorithms—a kind of attentional death in which a person is no longer overwhelmed because they have given up. You could also see it as an attempt to hold their footing as the zone floods with shit. Because everything is happening too much, too fast. More.

There is a cost to all of this—a flattening of every event, feeling, and piece of art, commerce, joy, and suffering into the same atomic unit of attention, all of them easily replaced by what comes next. The worst, most shameless people in the world already understand this and use that cold logic to their advantage. You do not need to justify a war if you believe that, ultimately, people will lose interest in it and move on to the next outrage.

I have suggested in the past that our information ecosystem is broken. But I now suspect that’s wrong: This is how it is meant to work. These online products sustain themselves by making us dependent on the content that makes us feel powerless and miserable. Where does this all lead? To further exploitation? To some kind of informational oblivion? Or will there be a breaking point, a moment when the addled masses reject the logic and speed of our information environment? I can’t say—but I’m monitoring the situation.

Why Is It So Hard to Make a Good Weather App?

2026-03-13T13:00:00-04:00

Subscribe here: Apple Podcasts | Spotify | YouTube | Overcast | Pocket Casts

How are we still getting caught in the rain? This week’s Galaxy Brain explores the world of weather forecasting—specifically the apps on our phones that we have come to rely on. As climate change intensifies storms and smartphones put hyperlocal forecasts in our pockets, we’ve never had more meteorological data. And yet plenty of people lament that their weather apps can’t get it right. Charlie digs into why we obsessively refresh our weather apps, why we blame them when they’re wrong, and what it really means to forecast an inherently chaotic atmosphere.

Charlie talks with the physicist Adam Grossman, a co-creator of the cult-favorite weather app Dark Sky that redefined minute-by-minute forecasting before being acquired by Apple. Grossman pulls back the curtain on how weather predictions are made—a process that includes government satellites, weather balloons, massive physics simulations, and machine-learning models—and explains why forecasts are improving even if it doesn’t always feel that way.

The following is a transcript of the episode:

Adam Grossman: It’s sort of the realization that all weather forecasts are going to be wrong, right? There’s nothing you can do about it. The key is: How do you convey that uncertainty?

[Music]

Charlie Warzel: I’m Charlie Warzel, and this is Galaxy Brain, a show where today we are going to get to the bottom of a question plaguing mankind since time immemorial: Do weather apps suck?

People have very strange relationships to weather apps. They check them obsessively, they love them, they talk about them, they pay money for them, and at the same time, they constantly complain about them. Weather apps often leave us high and dry or low and wet—whatever you want to call it. Weather apps are a feature of life, and yet the weather is super unpredictable. And so we get [these] tortured relationships with these devices, and they tend to be really, really important. As the climate gets more and more erratic, as there’s more instances of extreme weather, and as we become, increasingly, information junkies, we rely on these apps more and more. And frankly, a lot of times they don’t work the way we want to.

And so I wanted to demystify these weather apps. I wanted to talk to somebody who could tell me how they work, how they’ve gotten better, how they’ve gotten worse, whether we need all the information about the weather that we have. Remember, back in the day, we just used to look in the newspaper and get one forecast, or go to the local news and get a forecast in the morning and a forecast at night. Now we have all this information. What are we doing with it?

And so my guest today is Adam Grossman. Adam is a physicist who created the app Dark Sky back in the early 2010s, and that app quickly became an absolute cult favorite. It launched in 2012, and then Apple bought Dark Sky right around the pandemic and integrated it into their massive weather app. Adam then helped build WeatherKit at Apple for a long time, and he left to build a new app called Acme Weather, based all around this idea of trying to give people more access to more information and also communicate more uncertainty about the weather.

And so I thought Adam would be the perfect person to talk about this. He has this inside view of this platform, and he can help answer these questions: Why do we need all this information? Can we ever get a perfect, definitive forecast? Do weather apps suck, or do the users just simply expect too much from it?

Adam and I got to the bottom of all of this. Here’s our conversation.

[Music]

Warzel: Adam, welcome to Galaxy Brain.

Grossman: Thanks for having me. It’s exciting.

Warzel: So I want to start at the beginning here. How did you get into this job? This is an interesting gig, building weather apps, so have you always been a weather nerd? What is the background here? Walk me through that.

Grossman: So my background actually isn’t in weather. I have a physics degree. I ended up doing a lot of just software development, web development. But I think everyone is kind of a weather nerd, to a certain degree. Everyone gets kind of excited about the weather. I don’t know if it’s just built into humans in general.

I started doing weather probably 15 years ago. I guess it was in the summer of 2010. My now-wife, my girlfriend at the time, we were driving to Cleveland to go on vacation, because people go to Cleveland on vacation, evidently.

Warzel: I’m from Cleveland—I get it. I get it.

Grossman: Oh, are you?

Warzel: Yeah.

Grossman: Well, there you go. I’m in Connecticut, and so we were just driving west. And we pulled off at a rest area, and it was just a torrential downpour. Just cars on the highway were going 10 miles an hour. It was just a mess. And I remember opening up whatever weather app I had in 2010—I don’t remember what it was—and I looked at it to see, Okay, when can we go back out to our car and continue driving? And the weather app said something like, Seventy percent chance of rain. It was like, This is not useful, right? It was a torrential downpour. And then I went to the radar, and you could see it on the radar. And I remember thinking the whole time is like, How can we do this better? Right? If there’s rain right there, your app shouldn’t just say, Seventy percent chance of rain. It shouldn’t just say, It’s raining, right? It should be, Rain is gonna stop in 12 minutes, or whatever.

I started just thinking about the weather then, started playing around with just radar data, trying to see: Can we do machine learning, can we do computer vision to try to figure out where these storms are headed, right? ’Cause if you look at a radar map, you hit the little “Play” button, you see the radar moving at time. You know your brain can parse this as moving through time.

Warzel: Right.

Grossman: A computer should be able to do this, right? So I built a little gizmo for just trying to predict just the next few minutes, right, up to an hour of what the rain’s gonna do, minute by minute, so that if it looks like it’s raining outside or you get stuck on the highway going to Cleveland, you can say, Okay, in 15 minutes, you can go out to your car. It ended up working, and then we decided, Hey, let’s do a Kickstarter to see if we can make an iOS app. And that was an app called Dark Sky.

Originally, Dark Sky wasn’t really a general-purpose weather app. It literally just told you what the rain was gonna do in the next hour. It didn’t even have temperature, nothing. And we always promised ourselves, We’re not gonna make a general-purpose weather app. There’s so many of those. That did not last long, because we realized people don’t want two weather apps.

And then in 2020, just as the pandemic was hitting, we ended up joining Apple to work on Apple Weather. And then four years later, a few of us left, and then shortly thereafter, we started Acme Weather, which is our new app and our new weather service.

Warzel: So I wanna get there, but I think it’s really important that you had kind of the canonical normal weather app experience, right, which is: You open the thing up; you see that it’s not reflecting your reality. Obviously, you had the means and the tools to do that—to change it, to make something different. But I think let’s just start very basic here.

I want to get into the nuts and bolts of how weather apps and forecasting works for the layman, ’cause I think it’s really important to foreground that for the rest of this discussion, which is gonna get into why these apps succeed sometimes, fail other times. But can you just walk me through—explain it like I’m 5—how do these weather apps work? Or how does weather—let’s start even there: How does weather forecasting work?

Grossman: Yeah, so it depends what kind of forecasting you’re talking about. So the one I just mentioned, which Dark Sky did originally, was very short term, very hyperlocal: This is exactly what’s happening at your location over the next few minutes. But that technique is not gonna work for what’s happening this weekend, right, or multiple days ahead. And also, when you talk about just climate forecasting, long-term climate trends, that’s a very different kind of forecast.

When people think about weather forecasting, they sort of think hour by hour, out 10 days, right? And that’s sort of the starting point, and then you could tack on other things. But the way that works, there’s sort of a pipeline. And the beginning of the pipeline is gathering a whole bunch of weather data. And by the way, the beginning of this pipeline is mostly done by government agencies, government weather services.

And so the first step is: If you wanna predict the weather, you gotta know what the weather is doing right now, right? You gotta know what the sort of initial state of the world is. And that comes from satellite data. It comes from weather balloons that they put up. So the National Weather Service puts up hundreds of weather balloons—I think a couple hundred every day. Weather balloons are nice ’cause it gives you sort of a 3-D slice through the atmosphere, so it’s temperature, pressure, humidity, things like that, but at different elevations, and that’s really useful for subsequently simulating the weather. There’s ground stations, weather stations, right? There’s buoys out in the ocean that measure things like water temperature and all that. And so you have all this data that gets collected.

And then that all gets fed into numerical weather prediction models. And again, these are generally models that are run by government agencies. And what they’re literally doing is just calculating the physics. It’s basically running a physics simulation of the atmosphere, given the initial conditions that you have. And they run these things on enormous supercomputers, and you get an output. They’re now starting to do things like using machine learning and AI to do the same thing, but dramatically faster.

Warzel: Let’s go to Dark Sky. You created this to solve, as you put it, a very real and kind of isolated problem, right? To me, as an outsider, I feel like, when you guys started to blow up, the push notification part of it was really important, right? Like, I have this app that is not only gonna tell me this thing, but it’s gonna reach out, make use of, then, what was sort of a relatively new thing—push notifications were somewhat novel at that time—and to say, Hey. Hi. This is gonna happen. Be aware of this. Grab the umbrella, or whatever. What did you guys feel like you solved that led to really blowing up there as an app?

Grossman: I think the big difference is Dark Sky was a weather app and weather service designed for your phone. I started it in 2010. We didn’t launch ’til, really, at the end of 2011, beginning of 2012. But these phones that everyone has with them at all times, these smartphones, they had not been around for that long, right? We have these always-connected internet supercomputers in our pockets, and they were pretty new then. And before that, something like Dark Sky just doesn’t make sense, ’cause you have no way to actually get the information, right? It’s very specific on where you are right now. And that’s not how people got their weather information, in their weather forecast before phones, right? They got ’em from your TV meteorologist, right, or the newspaper. And those necessarily have to be for broad areas, right, for your city, your part of the state. And so the type of forecasts that they would provide were for your region.

It wasn’t that I think we were doing anything super technically magical. It was the fact that we were tailoring it for your smartphone, and I think we sort of got ahead of the other weather services out there, who were still sort of thinking about it the old way, right? The forecast that you would give on the evening news, they just put that on your phone, and it was the same forecast, right? And I think it was the realization that you can do some fundamentally different things once you have an always-connected, always-on device.

Warzel: I wrote a couple of years ago about weather apps in general—and we’ll get into this a little more later, I think—the tortured relationship that a lot of people have with them. There was a stat that I pulled from this website called ForecastAdvisor that was just talking about Dark Sky in general during the time that it was up: It accurately predicted the high temperature in my zip code only 39 percent of the time. Do you feel like there were a lot of limitations to what you guys could do there?

Grossman: It was a lot of fumbling around, right? So, again, we just started with one specific kind of forecast. When we first started looking into doing longer-range forecasts, oh boy, we were very naive. We thought, Oh, you just go out and get the data and then plunk whatever the data says into the app, and you’re done, right? Yeah, it’s not like that, right? And so it takes a lot of work to try to figure out how to take this data and turn it into something that’s as accurate as you can be, right?

Warzel: Were there any big fumbles, where you guys did something and were like, Oh no, we didn’t mean to do that, or This does not work—abort. How does that work?

Grossman: Ah, man. Doing something like this, it’s pretty much all little fumbles, right?

Warzel: Okay.

Grossman: You open the app, forecasts are one thing, but also, what’s happening right now? What’s the temperature outside right now? Models will give you that—not necessarily great at that, again, ’cause of, like, microclimate effects.

I was just looking at station data now, and we’re here, and it was in the 30s, but sometimes there’s stations that’ll say it’s 78 degrees or –100, right? There’s conversion issues. We had so many times where our forecast would just be off by, like, a hundred degrees because of a faulty ground station that we were just trusting.

Most big data problems is 90 percent sanitizing the data, munging the data. And same with weather, right, is I think most of our issues come from just the data is weird in some way that we could have caught, but we didn’t catch it, because we were young and naive.

Warzel: How did you guys get better at that? Is that just simply the process of trial and error? Were there certain light-bulb moments down the road there?

Grossman: I think the biggest thing is—so we didn’t just have a weather app, right? We had a weather app and a weather service to make forecasts, right? And so most indie weather apps, they work on the UI, and they call to a third-party weather service. So Apple has WeatherKit, which is their weather service that developers who make weather apps can tap into. I feel strongly that to make the best weather app, you should have your own weather service—for many reasons, and we can get into ’em, but a big one is because you’re going to get a lot of complaints and emails from users who paid you for this weather app and your forecast was wrong, and by far, user complaints are the No. 1 way that we learn about problems and then go and learn how to fix the problems.

But there’s no light-bulb moment. There’s a million problems, and it’s always something different. And so we wait for really angry customers to email us and say we ruined their wedding because it rained when we said it wasn’t going to. And then we go back and we try to figure out what the commonality between these complaints are and see what we could do to fix it.

Warzel: Do you have a funny or good example of a reader thing where you guys were just like, Oh no, oh geez, something that stands out to you?

Grossman: It’s things like ruining people’s day, and that makes us really sad. In Dark Sky days—well, according to the user—we ruined their wedding ’cause we botched a forecast, and it makes you feel bad. People tend to not email you when you get things right, right? No one’s just like, Good job. Go you. You kind of need thick skin, but it is super useful, right?

So in Acme right now, we have a “Community Reports” section of the app, where people can submit what the weather actually is outside. You make a report. You can see it on the map. You can see everyone else in your area’s reports on the map. That’s useful for a couple things, but one of the things is it gives us real-world data of what people are actually saying so that we can then look at that and say, Does this actually match our forecast? If it doesn’t, why doesn’t it? Right? There’s always gonna be noise, right? There’s always gonna be error, but are there systematic things we can catch?

The other nice thing about it is the weather forecast is always gonna be wrong, and so it’s kind of nice to have that ground truth from other users in your area that are like a sanity check, right? You can turn on notifications for that, so if multiple people say, Hey, it’s raining, you’ll get a notification, if you turn it on, that, Hey, other people in your area are saying it’s raining, right? And so I think that helps us get around just the inherent lack of certainty in a forecast.

Warzel: Let’s talk a little bit about going to Apple. You guys go in there, and the Apple Weather app is ... So many people have [smartphones], and so many people default to the one that’s on there. That—

Grossman: I don’t think I’m allowed to say how many users. It is a crap-ton of users.

Warzel: (Laughs.)

Grossman: It’s amazing how many users use Apple. It is scary. Working at Apple on Apple, it is very scary ’cause it’s a ton of users.

Warzel: Was that just an unbelievable amount of pressure to be in there? What were you Dark Sky guys doing in there, specifically? And then, secondarily, was it just like, Oh crap, the stakes are so high right now.

Grossman: Yeah, so Apple always had a weather app, right, and they just used third-party data. Apple decided—this is an important app for people. Apple’s really big on sort of owning the technology that powers their ecosystem. And so they decided, We need to have a weather service; we need to have that capability in-house, so that they can do all the things that they wanna do. They don’t have to be reliant on a third party. And so that’s why they brought us in, was to work on that. And that turned into WeatherKit, which is, again, it’s the behind the scenes [application programming interface] API for developers to deliver weather forecasts. And that’s what the Apple Weather app uses, so Apple Weather uses the same WeatherKit that, if you’re an iOS developer and you wanna make your own weather app, you would use WeatherKit for that. And so that was what we did, was come in there and work on WeatherKit.

Again, it’s kind of scary going from sort of a very niche, small, tiny company with what we thought were a lot of users, but not compared to Apple, and then going to this giant company. Yeah, it was a little stressful.

Warzel: What was the reason to leave and start Acme? What prompted that?

Grossman: So I have been a huge Apple fanboy ever since I was a tiny little kid. Getting to go to Apple and work on that was, for me, a dream come true. It was just absolutely amazing. Everyone there was great. The problem is, it’s a giant company, right? So you go from, like, the smallest company in the world, where you could just do whatever you want, and then you go to an enormous company, where there’s a ton of stakeholders, right—you can’t do whatever you want. Myself and the other Dark Sky people just found that we missed the small, scrappy start-up days at Dark Sky where you could come up with a crazy idea one day, work on it the next couple days, and then just ship it out. And if something breaks or people don’t like it, you can go and you could fix it and you can iterate. I think we just missed that, right? And so it’s just not something you can do at a big company, whether it’s Apple or anyone else.

Warzel: What do you, or did you—maybe it’s the same—see as the current hole in the market right now for weather apps?

Grossman: When I left Apple, and there are a few of us at Dark Sky who ended up leaving around the same time, I don’t think we thought we’d get back into the weather business again. But then it’s kind of hard, having done it for so many years and then having to use someone else’s weather app, right? It’s just like, Oh, but I want the weather app to do this, right? It’s like, Why are they doing it this way? I wanted to do it this way. And so we ended up just getting frustrated with the existing weather apps. And so our focus at Acme is—it’s sort of the realization that all weather forecasts are going to be wrong, right? There’s nothing you can do about it. The key is: How do you convey that uncertainty?

My favorite UI for weather, by far, is your TV meteorologist. You watch her; she says, Hey, there’s a storm coming in, but the European model has it being pushed up to the north, and so maybe instead of snow, we’ll get rain in the afternoon. They convey the uncertainty. They tell you what may or may not happen. And I think that makes a huge difference, especially for storms, right? Pretty much every weather app on the market just says, Hey, here’s what we think is going to happen. And this is our best guess. It’s, “How do you convey that uncertainty, and how do you deal with it?,” I think, is what was lacking in a lot of weather apps, and that’s sort of our focus with Acme.

Warzel: I wanna dig a little more on this with the current state of weather apps. And as someone who’s made them, how has the need for information changed, you think, over the last decade, decade and a half, as weather has gotten more extreme? Is it just that people are just more information-hungry, you think, now than they were, or do you think that there’s actually a genuine need, given the rise of more unpredictable or extreme weather?

Grossman: It’s probably both, right? And it’s not so much that it’s more unpredictable—actually, weather prediction has been improving faster than the weather has [become] more chaotic, so weather forecasts are getting better over time. Everyone listening to this is probably going to complain and say, My weather app sucks; it’s not getting better, but statistically, they are getting better.

But yet, to the extent that there are more just things that impact your day, people are just sort of more demanding now, right? Again, you used to watch the weather—you’d read it in the paper in the morning and then watch it at night on the news and then hope for the best, right? And so I think, now that everyone has weather apps on their phone, I think they’re more demanding for the information that they need right now or in the immediate future, right, definitely, because I think people are checking it way more often than they used to.

Warzel: So three years ago, I spoke to this weather-forecasting consultant and he told me, “The general public has access to more weather information than ever, and I’d posit [that] that’s a bad thing.”

Grossman: (Laughs.)

Warzel: Agree or disagree?

Grossman: No, I—well, I don’t know the context in which he said that. But no, more information’s always better than less information, I think, right?

Information overload is definitely a thing, right? And so weather apps used to be very simple. It was just what are the current conditions and then maybe, like, an icon and temperature for the next 10-day. And now people are demanding more than that, right? And it’s not that having that extra information is bad. It just makes it more challenging on what do you do with that information, right? How do you convey that in a way that isn’t information overload? That’s really on the people making the UIs and presenting that data, right? I think the demand for more data is, I think, totally legitimate. If that data exists, give it to me and give it to me in a way that I can understand it, I think, is the way to go.

Warzel: The context of that quote was this thing that you had just said a minute ago, right, where people are like, Just ask why they suck, right, why weather apps suck.

Grossman: (Laughs.)

Warzel: And I’m like, Do they suck? I am a little bit frustrated on your behalf about this because it’s like—

Grossman: They do not suck. They are wrong sometimes, but I guess it depends on what you mean by “suck.” You can get into the statistics of it and be like, Okay, what’s the Brier score for your precipitation probabilities?, right, and you can measure things.

I think that, yes, always having your weather on you at all times does make it more obvious when it’s wrong. I think we notice way more when it’s wrong than when it’s right, right? When it’s right, it’s just like, Okay, of course it should be doing what it should be doing. When it’s wrong is when you get mad, right? And that’s what you remember.

So, yeah, I don’t know. They don’t suck. They’re getting better, slowly. Forecasting is getting better. But we contrast that with people are checking it way more often. If you’re just doing tick marks on how often it’s wrong, you’re gonna have a lot more tick marks now just because you’re checking it way more often.

Warzel: I sort of agree with this. I think that it’s that people want certainty; they want definitive. And I think this is just the way that things are right now, right? We are in a moment of low trust, right? Just broadly speaking, in the world—I work in news. It’s a moment of relatively low trust of institutions of all kinds, right? They want something definitive when things feel uncertain. And I think, at the core, nobody can offer a truly definitive thing. Do you agree with that?

Grossman: People would love certainty, but I think what they’re really after is, if it’s uncertain, they wanna know that, right? It’s sort of, What is your certainty around your certainty of your forecast? Right? And I think that’s what people really want. If there’s a storm incoming and it’s just different models are saying different things, it’s very different for a weather app to just make a guess and be like, Okay, I’m just gonna go with this and give this. It’s a very different thing to say, Okay, look, the forecast is uncertain now. Here’s what might happen. Here’s how you can prepare yourself. Same amount of certainty in both cases, but being able to actually convey and tell people, We are uncertain, is, I think, a form of certainty—I’m trying to figure out what the right word is, right? I think people want that information. If it’s uncertain, they wanna know that it’s uncertain.

Warzel: You said that these forecasts are getting better. You mentioned machine learning and artificial intelligence. What, as you see it, is the impact right now of AI? Is AI actually making these forecasts better? Is it giving you, as someone who’s running their own service, more opportunities to crunch the data better, organize it, present it? What is the generative AI stuff doing for you right now, as someone building this?

Grossman: Yeah, so there’s different places to insert AI and machine learning into the forecasting. The big one is using it to do these numerical simulations of the atmosphere. The benefit there isn’t outright getting better forecasts. The real benefit is, it’s computationally orders of magnitude more efficient and faster to run a forecast. Doing the physics is just ludicrously expensive, and AI can do it at a minuscule fraction of the cost. And what that gives you is (a) you can run these much more frequently. So something like [Global Forecast System] GFS, which [at] the National Weather Service, [is] their global model, that updates four times a day, right? If, with AI, you could do it once an hour or once every half hour, you could get much more rapid updates, which is important for things like extreme weather, right? If you have a storm coming through the Midwest and it could spawn tornadoes, you want the best, most up-to-date forecast you can, right? And so doing it faster is huge. Because it’s so much more efficient, you can do it at higher resolution—you can capture more of those microclimates and potentially get better forecasts just by doing it that way. And so I think that’s where AI is helping.

And I should note that when we say “AI” here, we don’t mean plugging in data to ChatGPT, right?

Warzel: ChatGPT, yeah, exactly.

Grossman: These are weather-specific machine-learning models. And so what we do is we take those models, the model outputs, and then we use machine learning to do things like microclimate adjustments so that we can take advantage of high-resolution terrain data to give you better forecasts. We do it for generating thunderstorm probabilities, precipitation probabilities, and so we train models to do that.

What’s, I think, really interesting—we haven’t done this yet, but I think generative AI, things like ChatGPT, might be able to help convey that information. Again, like I said, I think the best UI is your TV meteorologist, right? But maybe, with the new on-device models that are coming out, things like that, maybe it could figure out how best to convey that information, how to convey the uncertainty. If it knows who you are and what you care about and that you walk your dog every morning and every evening, maybe it can help you tailor the forecast for that. And I think that’s more speculative, but there’s different places where I think machine learning can slot in and it can help in each one of those steps to make it better.

Warzel: It sounds like the project of Acme Weather right now is, as we were talking about, to not just convey the uncertainty, in a way, but to build some of that trust, right to work through that. And something that this makes me think of, and, again, not to get overly political, but the government is what collects a lot of this data, right? There’s been a lot of change in the government, a lot of shake-ups around research, but also around funding, cuts to—

Grossman: Data collection, right? Satellite, earth science, yeah.

Warzel: Yeah, cuts to different government organizations that may or may not be collecting this information. Does that offer concerns for the quality of the forecast?

Grossman: Anytime projects and funding gets cut, there’s downsides to that, right? You don’t have as much data that you would otherwise have. Or maybe, with proper funding, you would have gotten new satellites that have new capabilities that can push forecasting further, and then you just end up not having that, right, and so the improvement in your forecasting isn’t where it needs to be.

That’s what I’m worried about, is things like that. I’m not worried about politicizing the data itself, right, ’cause I don’t think I see much of that, but I think the issue is, as funding gets cut, there’s less we can do, less data that we can collect.

Warzel: Do you feel like that makes your job more difficult, as someone who’s building one of these things, if there are those concerns just out there in the ether?

Grossman: It just adds uncertainty, right? I’m an optimist. think we’re gonna muddle through. I don’t envision NOAA just dropping all their weather forecasting, right? If so, we rely so much on their data collection and data from other organizations. I worry about it in the abstract, but I don’t think it affects our day-to-day yet, and again, fingers crossed.

Warzel: What would you say to the person who, again, this is similar to the Do weather apps suck? or whatever, and I’m not asking you to defend them, but in terms of the state of this particular slice of the weather industry, the weather apps, what’s your message to them right now? Is it “Trust us”? Is it “We’re getting better”? Is it “Tell us exactly what you need”?

Grossman: Yeah, well, that’s the thing about the weather space is, especially weather apps, is everyone has their platonic ideal of what they want their weather app to do and everyone’s idea is different. Our pitch is: If we’re wrong, we don’t wanna surprise you that we’re wrong, right? If we’re wrong and that’s surprising to you, then I think that’s a failure on our part, right? We wanna tell you if we think we’re going to be wrong so that if we are, you’re not like, Goddamn it, you ruined my wedding, right? I want to avoid that, right? And so I think that is what we’re striving for, is to not catch people off guard. But if you are and you think the weather sucks, please let us know because, again, that’s the best way to fix it, is for people to yell at us.

Warzel: Adam, this has been extremely eye-opening and informative, and I feel like I have a better handle than I did when we got into this conversation about what the heck’s going on when I pull to refresh on my phone. So thank you so much for this.

Grossman: Thank you. This was fun. Feel free to email me if the forecast is wrong.

[Music]

Warzel: That’s it for us here. Thank you again to my guest, Adam Grossman. You can email him, but please be nice if his weather forecast ruins your day. If you liked what you saw here, new episodes of Galaxy Brain drop every Friday, and you can subscribe to The Atlantic’s YouTube channel or on Apple or Spotify or wherever it is that you get your podcasts. And if you appreciated this work and you wanna support it and the work of all my other colleagues, you can subscribe to the publication at TheAtlantic.com/Listener. That’s TheAtlantic.com/Listener. Thanks so much, and I’ll see you on the internet.

Did Netflix Ruin Movies?

2026-03-06T13:00:00-05:00

Subscribe here: Apple Podcasts | Spotify | YouTube

Few companies have reshaped American culture as aggressively as Netflix. This week’s Galaxy Brain charts how we got here.

Charlie Warzel talks with Atlantic film critic David Sims about Netflix’s strange, sweeping arc: from red DVD envelopes to a streaming colossus with 325 million subscribers. Sims explains how Hollywood initially shrugged off streaming as a novelty, only to watch Netflix reshape both distribution and the aesthetics and economics of entertainment itself.

Together, they discuss the rise of binge culture, data-driven green-lighting, and the tension between prestige projects and “second screen” slop built for distracted viewers. The conversation also examines Netflix’s stance toward theaters, its aborted bid for Warner Bros. Discovery, and the deeper question haunting the industry: Has Netflix simply exploited technological inevitabilities—or has it rewired our expectations of what movies and television are supposed to be?

The following is a transcript of the episode:

David Sims: When Hulu and HBO and all the other streamers start to crop up later in the game, it’s kind of like: You have Netflix, and then maybe you try another one. But you’re not gonna let go of Netflix. Netflix had just already won the war.

[Music]

Charlie Warzel: I’m Charlie Warzel, and this is Galaxy Brain, a show where today we’re going to talk about red DVD envelopes, the streaming wars, and the company that upended Hollywood.

Awards season will wrap up soon this month with the Oscars, which means it’s a good time to talk about Hollywood. And you can’t talk about Hollywood without talking about Netflix.

It’s difficult to imagine a company that’s had a greater impact on the entertainment industry over the last two decades. Since its founding in the late ’90s, Netflix has continued to do one thing over and over again: use technology and the internet to exploit convenience and wind its way into our lives. First it was a website that allowed you to pick your favorite DVDs to be shipped to you in the mail. Then it launched into streaming, original programming, a full movie studio. Now Netflix hosts live TV, award shows, sporting events—and is even a home for podcasts. The company has more than 325 million subscribers.

Netflix’s story follows the classic tech-company arc. The platform didn’t just disrupt how people watched movies and TV; it changed the culture and the fabric of entertainment altogether. Netflix has influenced the way that many movies look, feel, and sound— even how they’re conceived of and green-lit. The company has had its hand in creating everything: from auto-play, second-screen-binge mode-algo-slop to prestige award-bait projects. All of Hollywood’s hopes and anxieties—the decline of theatergoing, the data-driven writers’ rooms, you name it—Netflix sits at the center of all of it.

It’s a weird moment for the company. Back in December, Netflix made an offer to buy Warner Bros. Discovery in a deal worth approximately $82.7 billion. The purchase would have made Netflix arguably the world’s most powerful entertainment company. But Paramount Skydance, headed by David Ellison and backed in part by his father, the centibillionaire [co-]founder of Oracle, Larry Ellison fought the deal. Paramount Skydance submitted a revised offer to buy Warner at $111 billion. Netflix backed out of the deal last week. Some industry observers argued that Netflix dodged a bullet—or at least a lot of debt and regulatory headaches—by backing out. But now Netflix is at something of a crossroads.

And that’s why I’ve called on my colleague David Sims. David is a staff writer at The Atlantic, where he is our film critic and writes about the culture of entertainment. He’s also the host of the excellent podcast Blank Check. I wanted to talk to David about Netflix’s historical arc—how it became such a juggernaut and what it has done to transform Hollywood and all the ways that we consume entertainment. By all accounts, it feels like Netflix has won. Is that a good thing, a bad thing, or just inevitable? David joins me now to hash it out.

[Music]

Warzel: David Sims, welcome to Galaxy Brain.

David Sims: Hi, Charlie; thanks for having me.

Warzel: We’re approaching the terminus of award season and the Oscars. We also just had a lot of news around Netflix, Warner Bros., Paramount. Media consolidation. Growth hellscape/landscape, etc. So I wanted to have a conversation about Netflix, broadly—Netflix’s impact on Hollywood, on the industry, on all of us. And our eyeballs and our fragile little primate brains. So I thought it would be great to just start off very, very quickly: What is your first memory of Netflix? Your first Netflix experiment?

Sims: I feel like it gets referenced in The O.C. in maybe 2005.

[Clip plays from The O.C.]

Sims: That was the first time that I was kind of like: This is breaking containment. Like, you know, regular people are doing this; this is getting referenced. Like, people know about getting your discs in the mail. You gotta remember just—and I feel like this is almost forgotten now—DVDs were so vital to the sort of ecosystem of Hollywood, right? Like home video, for years, had been this sort of profit, you know, add-on. And that was fine.

And then DVDs come out, and it basically meant that you could make the worst movie of all time—kind of bomb in the box office, kind of not work out—and then you’re going to still make like 40 million extra dollars. Like, just from DVDs. it was a glorious era for Hollywood. And Netflix was just additive. Like, yes; all they were doing was buying discs and then sending them out to people in the mail. But it was just all part of, like, this wonderful cycle of extending a movie’s life and getting it out to more people. And yeah; as a college student, it was perfect for me. I got the three-disc plan. I don’t know if you did. Some people only did one. I would always have three. So I would have one disc that was a TV show, and then like one disc that was my next movie to watch, and one disc that I was sending off, like, you know, that I had just finished. That was sort of my Netflix cycle back in the day.

Warzel: I’m curious; do you feel like—and this is going to be a bit of a theme of the conversation, I think—did Netflix’s DVD business kill video stores? Or did it accelerate something that was already happening? They were already kind of on their way out. Like, how do you see that influence?

Sims: Netflix murdered Blockbuster in the way that Amazon killed Borders or whatever. Where it’s sort of like: Blockbuster had hastened its own demise; like, Blockbuster was ready to be killed. And It’s a bit of an urban legend. I think that, you know, it was a sort of one-to-one; like, Netflix came in and Blockbuster ended. But you know, it was just kind of like: “The internet is here; people want to pick their movies on a computer.” Netflix let them do that.

Warzel: I’m conflicted, because in one sense, I refuse to listen to any Blockbuster slander in any capacity. But, I’m just kidding.

Sims: Well, you had me on your show, so I’m going to slander the hell out of Blockbuster. Sorry. Yes, go ahead.

Warzel: But also, at the same time, I think this is a pattern with tech disruption and also with Netflix throughout the history of the company. Which is this idea of, like: Did Netflix accelerate certain things? What is Netflix responsible for? Which parts of the changes in Hollywood is Netflix responsible for? But what I want to get to is—so we have the DVD. And then Netflix decides to launch this streaming service. Right. And what I found in researching this, that I enjoyed, is: It launches a streaming service. Company’s stock drops six percent on that. And it seems like there was, at the moment, a little bit of like: This is a really stupid idea. You guys have it all with this physical media.

Of course, looking at that now, that seems kind of ridiculous. I’m curious, from your perspective, if you can walk me through a little bit about how Hollywood reacted to all of that. Right? Like, how the early days of streaming, how that kind of changed the industry. Or how people were thinking about that inside Hollywood.

Sims: I think they weren’t thinking about it much at all. Like, every time streaming stuff happens—in the sort of narrative of TV, Hollywood, movies, whatever—it catches them completely off guard, and they have no concept of it as anything but a novelty. So the whole thing with Netflix: When it starts up the streaming site, I mean, I remember looking at it in 2007 on my cruddy Dell laptop, that I’m sure would start wheezing and issuing steam if I tried to stream a movie on it. But it was crazy, because every movie was available. Because every studio was like, “Yeah, sure, you can have our entire second-run library. That’s fine. What do we care? Do what you want. Like, you know, how many people can even use this service?” And there was this sort of brief, kind of free-for-all, just like Wild West-y feeling to the streaming stuff. Because Hollywood was like, “This is how we make our money. The movie comes out in theaters. That makes us money. We put it on home video. That makes us more money. And then we we sell it to cable TV. That makes us more money.” Netflix is like, “That’s a little bit of extra garnish for us.” Like, who cares?

Warzel: In that sense, you have these companies not knowing what is going on. Is Netflix in this moment just gathering this information? Because I remember when the streaming thing happened, my first experience with it was … I think it was around Lost, right? Like, I had not watched Lost on actual cable at the time. So I was, whatever, three-and-a-half seasons behind. And I experienced the binge phenomenon myself. Was Netflix, at that time, just learning like, “Okay, all these fools have let us just have access to this content”? And we now realize, like, the people will watch as much as they physically can with their minds. This is like—did the data-collection stuff start at the beginning, you think? With them always being savvy?

Sims: It’s—how much do you want to buy into the sort of Netflix myth? I remember when they had a competition for someone to design a better algorithm than the one they had, right? Like, and this is pre-streaming. This is back when it was disc rentals. But they were like, “Hey, if you can beat our recommendations engine, we’ll give you like a million dollars.” And I think somebody did this. You know, like some coder that suggests that. Certainly, of course, as their business is taking off, they start to realize like, Right, the most important thing for us is to figure out what people want, and how to steer them toward what they want. And how to then, you know, turn that into much more profit for us.

Warzel: So when did, in your mind, Hollywood … when did they catch up to this? What was the moment when they realized This is bad news to have just be giving all this stuff out?

Sims: Two questions. Those are two questions, right? Because the first one, I think it’s around 2010. Warner Bros. like signed a huge deal with them to stream stuff. So that I think is when Hollywood is like, “This is a big deal. And it’s great news for us. We get money; real money. You know, we’re gonna start to make real money licensing stuff to streaming.” When do they realize that Netflix is going to get into their business and essentially, you know, start cannibalizing their business? My guess is that’s probably more sort of—I mean, you could say 2013, which is when [Netflix launched] House of Cards. But I feel like even that was a little scene, a little bit of a novelty.

And it’s not for another year or two that it starts to get a little more freaky, [with] the idea of a Netflix movie being treated like a real movie, even though it didn’t play in a theater. Which is the sort of core existential nightmare that many people in Hollywood still struggle with.

Warzel: So this leads to—right, this is what kicks off the streaming wars. The sort of the golden age of all of that. And I feel like …

Sims: Golden is a pretty loaded word to use for that. But sure; yes. The streaming age.

Warzel: Well, “golden age” in terms of, I guess, like green-lighting shows, right? And like, this notion of competition. Of, you know, “We need to program these things with new original stuff.” And whether that stuff is algorithmic fodder based off of, you know, what people will watch, or if it’s prestige stuff—I’m curious how you see this time. This mid-2010s time. Because it feels like, simultaneously, there is all this money flowing in; there’s all this stuff getting green-lit. It feels like this moment where, you know, Hollywood’s really grappling with people not going to the theaters in the same way. How do you see that moment? Was it this period of, like, “This is good in some ways”? Or was it feeling like, “This is just degrading the art”?

Sims: So, when’s Beasts of No Nation? That’s 2015. So that’s the first Netflix movie that, you know—it was a serious movie that they tried to get awards, and all that stuff. And I guess it all happens kind of fast. So what’s happening in Hollywood in the 2010s is: Marvel has sort of distracted them all. In terms of, like—every studio starts to panic that what they need is a gigantic sort of never-ending franchise that they can pump out three editions of a year. And that can be the sort of temples that they build everything else around. And I feel like their eye gets taken off of the streaming. So Netflix sort of starts to rush in to fill the more midsize movie space and TV space and everything like that. And obviously, the thing that they couldn’t really have predicted—or it would have been hard for them to spool up as quickly as Netflix does—is: Netflix becomes like a utility. Like, everyone has Netflix, right? So when Hulu and HBO and all the other streamers start to crop up later in the game, it’s kind of like: You have Netflix, and then maybe you try another one. But you’re not gonna let go of Netflix. Netflix had just already won the war.

The only reason I objected to using the golden—like, to me, the golden age of TV is what we were just talking about. The era that Netflix launched out of. Which is simultaneously the sort of HBO, the prestige cable, all that stuff—like The Sopranos, Deadwood, The Wire, all those shows—and then the sort of glitzy network stuff like Lost and Grey’s Anatomy and what have you, and The O.C., that create the binge-watching thing. And so that’s the golden age. And then the streaming age is what comes after. Which is where Netflix takes the reins, and they start making the content. And now it’s not actually the TV we liked. It’s TV that’s sort of designed to be binged. It’s designed to be a little easier to watch if you’re distracted.

Warzel: Right.

Sims: And the answer of why it feels like this is: because this is what it is. Movie scripts have been stretched to 10 episodes. Because people start pitching movies, and everyone’s like, “Movies; eh. If your movie doesn’t have a superhero, we don’t care about it. Could it be a TV show?” And so you start to see lots of things get turned into streaming TV that maybe, you know, didn’t have enough plot to fill 10 episodes. But if it’s on Netflix, people will watch it. So that becomes what we’re all dealing with.

Warzel: It’s really interesting covering tech and watching the ways that things become so recursive, right? Like, it’s basically: You get a thing that’s great. And then as technology interacts with it, or works on it, what you get is like a game of telephone—with that same thing that is just, you know, degraded.

Sims: Right.

Warzel: And I think that brings us to what you have been talking about: this rise of the algorithmic-friendly entertainment. The ambient-viewing stuff, the big dumb titles. Like, I saw a Guardian article about Netflix that was talking about how the titles have become so incredibly obvious on some of the more trashy content. And one is just called Tall Girl, because it’s about a tall girl.

Sims: Hey man, Netflix’s tall girl. She’s tall! I mean, Hunting Wives. Hunting Wives was a huge hit for them last season. And that was a show that multiple people were like, “You should check out Hunting Wives.” Which I haven’t gotten to yet. But where you’re like, “Let me guess: It’s about some hunting wives. Wives who go hunting.” And like, you know, they’re so focused on “Can it be sold?” You know, in the carousel, right? Like, “Can you basically design me a sort of punchy image and a quick title that’s gonna work as someone is scrolling through a hundred different opportunities?”

But then, what’s funny about Netflix is it will also produce stuff that’s really worthwhile. And you can tell when a lot of the sort of controls were waived, right? Where there was probably less studio notes of, like: “And make sure it works on a phone. And make sure people explain the plot five times during the episode.” You know, where an auteur who Netflix takes seriously is being given, you know, a lot of room to make a passion project. Or something like that. Because Netflix knows the occasional sort of awards-y bump is really helpful to the bigger experiment.

Warzel: So you have Netflix collecting all this data about what people are watching, what they’re doing. It leads to the creation of some of these shows like House of Cards, which end up being eminently watchable. Still feeding into that prestige, even if it is a little on maybe the slightly trashier side.

And what we see now, though, is this collection of more information to create these ambient-style shows. These second-screen-style shows, right? The shows that you’re just supposed to do doomscrolling, and you can watch people announce what they’re doing all the time in there. And I’m curious; is this like … has Netflix learned the wrong lessons from all of this? Like, they were using the data in service of something that is relatively high quality. [Is] the data just, like, degrading the brand and all of what they’re producing?

Sims: I guess the argument against “they learned the wrong lessons” is that Netflix is successful and profitable, and lots of people subscribe to it. And so, if you just follow the money, they’re right and we’re wrong. Right? But I feel like there’s no really good argument for why so much of their TV—especially TV—needs to be made in a way that almost assumes the viewer is not paying attention. Right? Like when I call it a second-screen show or whatever, if like someone’s basically scrolling Instagram while they have the Netflix show on in the background. So the Netflix show needs to be sort of obnoxiously, loudly plotted as possible—so people can kind of track what’s going on. It’s assuming the worst to your viewer. And so much of good TV assumes the best of its viewer, right? Like, assumes that viewers can pay attention and figure things out, and maybe talk to each other if they didn’t figure something out. And it sort of points to, I feel, like a lot of dissatisfaction a lot of people have with how TV is these days of like. Why is it like this? It’s like: Well, they’re kind of assuming the worst of you.

Warzel: This is maybe an unanswerable follow-up to that. But do you think people will just lap up all the slop for as long as possible? Is there a point where—and I feel this way about content everywhere, right? Like, are we going to reach a point where people are like, Just stop. Like, stop debasing me with this thing?

Sims: Oh my God. As, like, will people stop watching things? I don’t think so. You know—has the world broadened to the point of “a thousand points of light” versus, like, three networks of TV? You know, back in the ’50s and ’60s. Yes. So I suppose the answer is like, well: They’re always gonna have the choice to watch something else. But that’s why the game Netflix has played of “We need to have the biggest user base” has worked out for them. In terms of like—yeah, well, sure, maybe people want to watch something else. But more likely, they’re going to use the thing they pay for, because it’s the easiest option.

Warzel: I wanna see how many of the 10 most popular shows globally, of all time, in Netflix that you could guess before just totally giving up. What do you think is the most popular Netflix show?

Sims: Stranger Things. Is it Stranger Things? Either Stranger tThings or Squid Games, Squid Game. Whatever.

Warzel: It’s not. It’s neither.

Sims: Okay, what is it?

Warzel: It is Wednesday.

Sims: That would have been my third guess. Which is one of those things that you’re like, “Does anyone ever talk about Wednesday?” I know it was unambiguously a hit. It was unambiguously seen. But it’s not like you walk the streets hearing people go, “I can’t wait for more Wednesday.”

Warzel: Not only that, but that Season 1 is the most viewed. But Wednesday, Season 2, is the fifth most. So it’s like, it did have the staying power.

Sims: Right; more recent.

Warzel: But Stranger Things [Season] 4 comes in at No. 3. So it’s not even the second. And Squid Game is not No. 2. Do you have one more guess at what No. 2 might be?

Sims: Uhhh, is it Bridgerton?

Warzel: No.

Sims: Is it, huh. Because I’m sure there’s also, like, reality stuff I’m not considering. And it’s not Squid Game.

Warzel: I’ll give you a hint; it’s more prestigious than you might think.

Sims: More prestigious. Is it House of Cards?

Warzel: It’s Adolescence.

Sims: Adolescence. Yeah; wow. So that all speaks to something that is almost illogical. But I guess it’s just: Their audience has gotten so much bigger. That like, you’d think, Yeah, well, surely something like a legacy show for them, like House of Cards or Orange Is the New Black or whatever, built up the bigger audience. But no. Like when Adolescence was such a smash last year, it was playing to the most subscribers Netflix had in their history. So yeah; that makes sense.

Warzel: But it all just speaks to exactly what you said. Which is like, one of the things that Netflix has done is, like, divorced. It’s added to the weirdness of popular culture. One of my big hobby horses is “Nobody knows what anyone is doing, because of the internet.” No one knows what anyone’s watching. Everyone knows everyone’s opinions, but doesn’t really know if they actually believe them, or what’s happening. Like, Wednesday is such a good example of this. A true phenomenon. But that doesn’t really—it certainly penetrates popular culture, but not in the Seinfeld-ian, “What’s gonna happen on ER tonight?” kind of way. It’s just; it’s super weird. But yeah. I’m glad that it stumped you. I was gonna be mad at you.

Sims: You did a good job stumping me. It’s just like the reheated-nachos element of Wednesday is just: It is a great way to think about what Netflix brings to the table. And Netflix has made good television, and it’s made good movies. I’m not saying it hasn’t. But like, Tim Burton in his kind of twilight years directed a spin-off of Addams Family, that’s kind of a high-school drama with a murder mystery. It just sounds like something a Netflix algorithm came up with. And no wonder it was a smash hit for them. Like, that’s why they have these programs. That’s why they do the things that they do. But will anyone, to use a Bill Simmons–ism, like: Will anyone be like bouncing someone on their knee, telling them about Wednesday, Season 1, when they’re a grandpa? I don’t think so.

Warzel: So I’m curious with this. Like, the algorithm-friendly, big, dumb, potentially trashy entertainment stuff. You mentioned to some degree how it looks, too, right? Like this feeling that everything from, like, the palette to … it all feels very scroll–phone based. Or it doesn’t really matter, because it’s second-screen stuff. How much of this has seeped into modern filmmaking? Like, like broadly speaking, the Netflixification of it.

Sims: Yeah; I would say some. It’s sort of a larger crisis, or a larger sort of existential question in commercial art right now. Which is like: “Why does everything look this way?” Right? I was recently watching Michael Bay’s Transformers for reasons I can’t really explain.

Warzel: You don’t need to explain yourself.

Sims: And, you know, it’s a movie that has some coherence issues. And it’s a movie about robots that turn into cars, and all that. But I was just like: God, this looks so good. It’s so well lit. It’s so thoughtfully made, as much as it’s silly, visually. And now, is the reason that all the movies kind of look like “that”—by which I mean, sort of like they’re a little flat, they’re a little under lit, everything looks just a little staid—is that because of Netflix? Because things need to be viewable on multiple different, you know, phone, iPad, TV, cinema screens? Is it because some people blame the way visual effects work these days? They prefer, you know, less lighting? I’ve heard that. I have no idea. Some people say it’s because actors now have … like, they arrive at set with their own lighting portfolio, and you have to light them a certain way. So you’re not allowed to make artistic choices anymore. I’ve heard that as well. I don’t know. But I do think Netflix is kind of part of it, in terms of like—what’s the most crucial to them is that it can be viewed in many different formats. And so, if your movie is gonna go for an artier thing, that might not translate on a smaller screen. And that's not gonna be good for them.

Warzel: So I wanna, just given what we’ve talked…

Sims: I know that was a lot to throw at you there.

Warzel: No, honestly; it was great. It’s weird, because this is again, the theme of the conversation to me—which is like, how much of this is a thing?

Sims: How much is their fault, right? And how much is it just like: This is what’s happening; what can you do?

Warzel: Yeah; how much. Right. Yeah; not to just totally skip over all of it where I wanna get to. But part of it is like—if Netflix didn’t do this, if Netflix didn’t come around in this way, wouldn’t somebody else just have done this? It just feels like the system’s there. Netflix has exploited it, but it’s so hard to chicken-and-egg. Like, did the evil geniuses at Netflix do this to our beautiful boy of film, you know?

Sims: You cover tech, Charlie. So you can tell me. Because I think the answer is: Netflix is maybe an accelerant, or maybe more aggressive, you know. It’s like, I feel like this happens in tech a lot, but you would know better than me. That like—there’s the established companies that kind of rule something, like computing or telecommunications or whatever. And then a newer thing will come in that’s initially disruptive and then becomes a colossus on its own. And that’s what happened with Netflix. Where it’s like: You could imagine a world where Warner Bros. are the people who are first with a streaming platform, and they kind of set the tone. And it’s a little more conservative, because it’s an old legacy company that doesn’t want to rock the boat too much. But I do feel like often you need to have this kind of upstart company set a new tone, and then the, you know, slower conglomerates sort of struggle to catch up. So maybe Netflix is responsible in that way. But it’s also like: Someone’s gonna do that, right?

Warzel: So Netflix went through this big messy pursuit, as we mentioned earlier, of Warner Bros. Ultimately, Paramount comes in over the top, pays a ton of money, has the sort of the Ellison/Trump, you know, possible “greasing the wheels on getting this through” connection.

And Netflix gets a nice $2.8 billion termination fee for going through the whole thing. You wrote back in December that Netflix’s potential acquisition of Warner could spell doom for cinemas down the line. Outside of even that, it felt like creatives and people in Hollywood, people were speaking out and just saying, “That is not good.” Right? Like, they’re just scared of Netflix having this power, of this consolidation.

Sims: Yes.

Warzel: But I want to talk about that concern in that moment when it seemed like it was going to be Netflix’s game to win. Really, kind of “Netflix as the apocalyptic force” stuck with me. And so I’m curious. From that, what is the reputation of Netflix right now in Hollywood?

Sims: What’s sort of interesting about the last six months, since the Warner Bros. bidding war broke out, is that I feel like there’s been a slight softening on Netflix. Partly because Netflix really, really ran like a big political campaign within Hollywood to try and convince people: “We’re not the monsters you think we are.” And so what happens is obviously—Warner Bros., which is quite successful, the film-branch company, but it’s laden with debt. It’s being built to sell by David Zaslav for the last couple of years. And I think Zaslav expected Paramount, Universal, the other big studios to try and grab it. Essentially to be like, “We just need to get bigger. We need to get bigger to fight Disney, to fight Netflix.”

Then Netflix comes in, and, you know, it initially wins the bidding war. And everyone starts panicking, as you’re saying. Because Netflix has just been, philosophically, really like hostile to the idea of the sort of classic “Release it in theaters, let people enjoy it, and a few months later it can hit the internet” sort of strategy, that’s existed for a long time. And everyone immediately is just like, “Okay, that’s it. The biggest movie studio in America that’s not Disney, Warner Bros., is about to vanish from theaters. That will kill theaters.” Like, “That’s it.” Like, you cannot survive without their, like, 15 to 20 big movies a year.

And then you saw people panic so much that Sarandos actually started being like, “No, no, no, no; it’s fine. I’ll commit to a 45-day release window. I will honor all of these commitments. Warner Bros. is going to be its own thing.” To the point that people started believing it. We’ll never know if he was fully on board or not, because it looks like Paramount’s gonna get the company. But it was sort of interesting to watch, because I almost started to believe Ted. He’s spoken so disdainfully of the theater experience. But it was kind of this question of: Why would Netflix spend $80 billion buying a company that releases movies in theaters?

It’s like buying McDonald’s and not selling chicken nuggets or whatever. It’s just sort of like—it’s how this company makes money. Surely you’re not going to buy, you know, Warner Bros. just to sort of prove some point, right? I just never really understood why Netflix would want Warner Bros., in terms like—Netflix has just always been very kind of apart from the Qwikster nightmare that they did like a million years ago. They’ve always been very focused. They grow at the right scale. They know what they’re doing, and they know what the next step is. And this felt like a very weird next step for them to be taking.

Warzel: Do you think some of it is truly just about power? At a certain point, you just have to show that you’ve won. I’m fascinated, too, by this idea of being so hostile to the theater experience. Is that just because: That’s our DNA, baby. As soon as we started doing the DVDs and allowing you to bring them home, which is obviously how the company started, we have always just been protective of the home experience. And that’s just who we are. And we won’t stray from that?

Sims: Right.

Warzel: I don’t know. I feel like the only way that you could think about it, because of that, is this idea of just power, right? Of the tech mindset. Of just like: We have to scale. Scale is just—if you don’t scale, you die. And we have to find weird ways to scale now, because we are so big.

Sims: I think that’s part of it. I think it’s this, like never-ending need to grow. And Netflix buying, you know, Warner Bros. and HBO. That’s growth, baby. Like, especially them adding the HBO stuff. Like, no doubt, that’s something maybe that would indicate an even bigger future for them. And also, yes; you’re fighting off consolidation from other studios, which I guess would be designed to rival you.

Warzel: Yeah. And not to go all late-stage capitalism on this, but at the same time, it’s like: This is the logic of tech companies. This, like, hyper scale. But, you know, they have a subscriber ceiling—in the sense that you just can’t get everyone to come do this, right? Like, you do grow to this point. I mean, maybe you can raise the price forever, but at some point people will probably look around and say, “Okay; I am paying $60 a month for this thing. I don’t know that I really want it. Whatever.” Right? And they’re staring down the thing that always befalls these successful tech companies—expectation of the forever growth. So what do you think is next for Netflix now, post-Warner?

Sims: So Ted Sarandos has said in these postmortem interviews, like, “I think we can do more stuff with theaters.” Like, he’s really trying to push the change of tune. And the funniest outcome of all this would be Netflix being like, “We’re gonna operate a little bit more like an old-school prestige movie studio. We’ll still be a TV company, a streaming-TV company, and that’ll be our big profit engine. But we want to actually rebrand to something a little classier, because we see how much you guys freaked out last year.”

That could happen. It doesn’t strike me as the best way to growth, but like you said, I don’t really know how they could possibly grow. Like, they are the biggest.

Warzel: It’s an amazing head game to to play, right? If you’re always one step ahead and pushing people in a direction that makes them feel uncomfortable. Like, if that’s the Netflix legacy with Hollywood: make people do things they don’t feel comfortable doing. Creating this streaming paradigm in this way that everyone has to catch up to you—and as soon as they start to catch up to you, just say, “No, we’re actually going to do what you guys did. Thanks for spending all that money on completely changing your business around. We’re going to go back to that, you know, a year ahead of you.”

That would be kind of like the, I don’t know, the trolling school of business. But I kind of love it.

Sims: Yeah; I kind of love it too. I just feel like, you know, Ted Sarandos is on as much of a high as he’s ever been right now. In terms of like, Everyone’s mad at David Ellison right now, and everyone’s mad at David Zaslav, and people actually kind of aren’t mad at me. And Netflix not getting to buy Warner Bros., you know, prompted a week of articles of people being like, “You know, Netflix isn’t so bad.”

Warzel: The other elephant in the room is the pivot to the generative void. The “Let’s just try to churn out movies without having all these messy people being involved.”

Sims: That’s the evil version. And they can do the good and the evil. They could be like, “We’re going to make art movies. And also, by the way, we will have an AI channel that just shoots slop into your ears.” Like … they could do both, right?

Warzel: The good bet is both, for sure, I would imagine. Okay, so to kind of land the plane here, I want to just like get your assessment on this whole arc, right? Because I think it’s really easy to start adding all of this stuff up. The ambient entertainment, the “we must announce what characters are doing 25 times because no one’s actually paying attention.” You know, all that stuff.

Sims: Right.

Warzel: Netflix has, in one way, for two decades, exploited convenience. And that is not necessarily a bad thing, right? Like, I don’t think we have to clutch our pearls about that. Convenience is good sometimes. But there’s also been this complete and total impact on the industry at large. But also with us as consumers—what we expect out of entertainment now, how we want to watch it. Our viewing, our consumption habits have changed alongside the production habits of it. And so I’m just curious. You, as a critic—as a reporter on the culture of all of this—where do you kind of net out on it, right? If everyone right now is sort of like, “Maybe Netflix isn’t so bad,” where are you when you think about this arc? Where do you fall?

Sims: I am sorry to come off as an entertainment centrist, but that is basically what I am. I love that. Look, when I was a teenager, if there was an art movie that came out in theaters in limited release—you know, New York and L.A. and San Francisco and like Chicago—the chances of it reaching me before it hit home video were, if I didn’t live in some big city, were tiny. And I love that you can get a movie like that in front of people within a few months on the internet—Apple, all the Amazon rentals, like all that stuff—while also giving it a run in theaters. That, to me, seems like a great way to sort of preserve the medium that film critics like I love. Without annihilating it while embracing convenience.

What I’ve never understood about Netflix is why there needs to be sort of a monotheistic platform. Right? Where it’s like: You simply must ingest it the way we want you to ingest it. You have to binge the TV; you have to watch the movie at home. All that. Like, just make all things available to all people in all ways. Like, it’s a great way to get art into people’s eyes. Like, they can pick how they want to experience it. Is that so wrong? Am I so evil to just be, you know, wanting everything for everyone?

I don’t know. I think some of these companies are like, “No; it’s a competition, and we need to win it.” I’ve never really understood that. To me, it’s just like: Everyone should just try to make the best stuff they can and get it to people every way that they can. In exchange, they can get money. Which, you know, is kind of how the whole business thing is supposed to work.

It’s insecurity, I guess, is the best way to put it. Like, even though Netflix won, they’re still basically like: “Yeah, but how do we know you’re not gonna leave us tomorrow for Peacock? So we have to keep you on the line.”

Warzel: Great, great Peacock plug at the end there, David.

Sims: Yeah. Everyone sign up.

Warzel: David, thank you so much for trying to make sense of this.

Sims: I’m doing my best. Thank you, Charlie, for having me on.

[Music]

Warzel: That’s it for us here. Thank you again to my guest, David Sims. If you liked what you saw here, new episodes of Galaxy Brain drop every Friday. You can subscribe to The Atlantic’s YouTube channel or on Apple or Spotify or wherever it is that you get your podcasts. And if you wanna support this work and David’s work and the work of all my colleagues at The Atlantic, you can subscribe to the publication at TheAtlantic.com/Listener. That’s TheAtlantic.com/Listener. Thanks so much, and I’ll see you on the internet.

A Technology for a Low-Trust Society

2026-03-05T15:06:10-05:00

A few hours before Donald Trump gave his State of the Union address, Republican sources told the PBS correspondent Lisa Desjardins that the speech would break records. The president would speak for more than two hours, she reported on X, and one reliable source claimed he might ramble on for 180 minutes.

The post went viral. At about the same time, the market started to move on Kalshi, an online platform where people can invest money in the outcome of a given news event. (Don’t call it gambling.) Forecasts on “How long will Trump speak for at the State of the Union?” shot up by 10 minutes after Desjardins posted: Armed with what they perceived as insider information, users thought they could make a buck by accurately “predicting” the outcome of his speech.

But others speculated in a different direction. “They’re leaking a bunch of stuff about a super long speech and he’ll go about 2 minutes short of the supposed mark and everyone in the white house will make $200k on it,” one Bluesky user, ‪@danvogfan, posted a few hours after Desjardin’s post went viral. In other words, maybe the sources really did have good information—but they were throwing others off track to manipulate the market and profit for themselves.

Prediction markets such as Kalshi and Polymarket have ushered in a moment when anyone with access to exclusive information related to a major news event can do this, even as the platforms themselves prohibit market manipulation. Trump ultimately didn’t speak for as long as the sources had said: He ended after an hour and 47 minutes. Anyone who had bet according to the information that Desjardins had reported would have lost money. “We live in such a profound dystopia,” another popular Bluesky user wrote above @danvogfan’s post after the fact.

[Read: America is slow-walking into a Polymarket disaster]

We can’t say definitively that any insider trading has actually happened, though other suspicious incidents have occurred. In early January, one Polymarket user bet more than $30,000 on Venezuelan President Nicolás Maduro being ousted just hours before he was captured by the U.S. military. (The bet paid out $400,000 and led Representative Ritchie Torres to introduce a bill that would ban federal workers from using prediction markets.) Last month, Israeli authorities charged two people on suspicion of using classified information to bet on military operations on Polymarket. And this past weekend, an anonymous trader who goes by the name Magamyman made more than $550,000 on Polymarket by betting on the timing of U.S. and Israeli strikes on Iran and the fate of its supreme leader.

Welcome to the democratization of insider trading, brought to us by platforms that let people wager on election outcomes, sports, and “Taylor Swift pregnant before marriage?” The prediction markets frame bets as tradable “shares” that rise and fall like stocks, financializing every current event and piece of online ephemera and generating a pervasive hum of paranoia: The world as a hedge fund, where everything can be a derivative. Greed is good.

Prediction markets claim to harness the wisdom of crowds to provide reliable public data: Because people are putting real money behind their opinions, they are expressing what they actually believe is most likely to happen, which, according to the reasoning of these platforms, means that events will unfold accordingly. Many news organizations, and Substack, now have partnerships with prediction markets—the subtext being that they provide some kind of news-gathering function. Some users who distrust mainstream media turn to the markets in place of traditional journalism.

But in reality, prediction markets produce the opposite of accurate, unbiased information. They encourage anyone with an informational edge to use their knowledge for personal financial gain. In this way, prediction markets are the perfect technology for a low-trust society, simultaneously exploiting and reifying an environment in which believing the motives behind any person or action becomes harder.

Polymarket did not respond to my request for comment. In response to concerns about trading on the outcome of the Iran strikes, the company said that it aims to “create accurate, unbiased forecasts for the most important events to society.” Jack Such, a spokesperson for Kalshi, told me, “War markets put Americans at risk and have absolutely no place in prediction markets.” Unlike Polymarket, which technically operates outside of the United States, Kalshi is subject to U.S. government regulation. It does not allow bets on wars or assassinations, though it did host a vaguely worded market pertaining to whether Iranian Supreme Leader Ayatollah Ali Khamenei would be “out.” After he was killed in the conflict, Kalshi did not resolve the market to “Yes,” enraging some users.

[Read: Your phone is a slot machine]

“We share the concerns about war markets, death markets, and insider trading. We don’t allow any of these on Kalshi,” Such also said. “Having a market on whether or not the U.S. will enter a civil war is insane. Not all prediction markets are the same.” (Polymarket has indeed hosted such markets.)

But these differences, though relevant in specific instances, do not have much bearing on the larger problems that these platforms contribute to. Prediction markets are selling a philosophy: Tarek Mansour, Kalshi’s CEO, said that the company is “replacing debate, subjectivity, and talk with markets, accuracy, and truth,” and Polymarket’s CEO, Shayne Coplan, said his company is “the most accurate thing we have as mankind right now.”

On X, both Polymarket and Kalshi have their own accounts that act as news feeds, where they post engagement-baiting and occasionally misleading headlines and speculate about world events that, conveniently, one can bet on via their platforms. On Tuesday morning, Polymarket’s X account looked an awful lot like a news wire, posting, “BREAKING: Ken Paxton projected to win today’s Texas Republican Senate Primary” and putting his odds at 83 percent. Paxton ended up with about 40 percent of the vote, slightly less than his opponent John Cornyn, who was polling at 18 percent on Polymarket at the time of the post; the two will compete in a runoff election in May.

The markets also encourage a kind of meta-game. People are betting on outcomes, but they are also hedging with side bets. For example, this winter, the Polymarket entry titled “Will Jesus Christ return before 2027?” climbed from 1.8 percent betting yes to roughly 4 percent betting yes in one month. The bizarre spike made the rounds online before a perceptive X user noted that the real reason for the change was that Polymarket traders had created a secondary market to bet on whether the odds of Christ returning would climb above 5 percent. Those traders were then manipulating the original “Will Jesus Christ return before 2027?” market to try to make money on their secondary bets.

[Read: You’ve never seen Super Bowl betting like this before]

This means that the markets don’t always reflect what people think will happen as much as they reflect what people think other people think will happen. This certainly gives the lie to the promise of accurate, truthful information from these platforms, as do the suspected incidents of insider training. When a market is manipulated by people with exclusive information, it does not provide clear, actionable intelligence to everyone else. That’s because many of these markets come and go quickly. Somebody who was up at 2 a.m. and happened to be paying attention to Polymarket’s Iran-air-strike market may have been able to pick up on Magamyman’s big bet and gain a subtle informational edge, but it is absurd to compare this subtle signal to credible reporting or intelligence. Kalshi, at least, seems to realize this; Such told me that the platform “bans insider trading not only because it’s unfair, but also because it erodes trust.” (Insider trading is also illegal, a point that a spokesperson for the White House repeatedly pointed out to me, without addressing my questions about whether the administration has its own rules forbidding government workers from participating in prediction markets.)

So they’re specious forecasting tools. Yet the prediction markets are bad for another, much more obvious, corrosive reason: Beneath the veneer of forecasting, the platforms are funneling gamblers to markets to bet on human suffering and acts of war. The top market on Polymarket’s homepage as I wrote this sentence was “Will the Iranian regime fall by June 30?” More than $6.7 million has been wagered on it so far. Betting on geopolitics and military operations allows traders to profit off of death, and it transforms people, politics, death, trauma, everything into commodities. In the wake of the first strikes on Iran, Polymarket briefly allowed trades on when a nuclear weapon was likely to be detonated. Current events, no matter how heinous, become entertainment, a business plan, or both—what Jason Koebler of 404 Media recently dubbed a “depravity economy.”

As the depravity economy grows, it will break whatever trust we have left in one another: If prop bets spur athletes to play differently, this poses an existential threat to the integrity of live sports. If people believe that anonymous government insiders are profiting off of classified information, what reason is there to trust anything that the administration says? There’s a term called the liar’s dividend, which describes an information environment where mis- and disinformation such as deepfakes become so prevalent that anyone accused of doing something awful can simply use them to cast doubt on genuine evidence. Prediction markets offer an insider’s dividend, creating an environment where the prevalence of prediction markets and insider trading becomes great enough that everyone assumes a given decision was made to enrich those with an edge.

This is the central lie of prediction markets: They claim to get us closer to the truth but, in the end, they make us less certain about the world. But this erosion of trust is a feature, not a bug, for these platforms. A world where people are suspicious of every motive is a world where the cold logic of gambling feels more rational. A zero-trust society is one where the prediction markets’ dubious “wisdom of crowds” marketing seems extra appealing.

In this way, prediction markets are a system that justifies its own existence—a well-oiled machine chipping away at societal trust while offering a convenient solution to its own problem. The prediction markets have done what any savvy trader or firm might—they’ve hedged their bets. The house can’t lose.

What Do the People Building AI Believe?

2026-02-27T13:00:00-05:00

Subscribe here: Apple Podcasts | Spotify | YouTube

Silicon Valley runs on hype cycles, and the AI boom is generating a new one—part gold rush, part ideology, and part quasi-religious devotion to building an alien intelligence.

On this week’s “Galaxy Brain,” Charlie Warzel explores the culture of this boom with the writer Jasmine Sun, who’s been chronicling San Francisco’s AI scene. Sun describes what this moment feels like on the ground, including a subculture of massive salaries, and a weird pride in leaning into tech’s strangeness. Together, Warzel and Sun unpack two major factions shaping the industry: the AI “doomers,” and the accelerationists. The conversation also traces Silicon Valley’s rightward drift—the “founder mode” backlash against regulation and employee activism and the rise of “Trump style” provocation-first tech marketing. Finally, Sun and Warzel address the jagged reality of today’s models, which are brilliant at some tasks and weak at others.

The following is a transcript of the episode:

Jasmine Sun: The way that AI progresses is in these fits and starts, and it’s going to diffuse into our society quickly, but also incrementally. And I don’t really want to wait around until that moment that AGI shows up and we can all agree on it before we start to think about what that actually means for us.

[Music]

Charlie Warzel: Every tech revolution produces its own distinct culture. The vibe of Silicon Valley’s early computing years was, in part, countercultural. There were government contracts, yes, but the builders of that moment were also influenced by DIY publications like Stewart Brand’s Whole Earth Catalog, which sought to “change the world by establishing new, exemplary communities from which a corrupt mainstream might draw inspiration.”

The dot-com boom of the late ’90s and early 2000s was fueled by an optimism that was much more profit driven, but also buoyed by the novelty of the commercial internet.

“Carpet the world with cheap technology, and clever hands will put it to work in a thousand ways never before imagined,” Wired wrote describing the moment. “Moore’s law boiled down to one word: more. The more you have, the more you use. While traditional economics are driven by scarcity, the world created by the microchip is one of abundance.”

People saw the internet and felt certain it would change everything. Money flowed in aggressively. The big bets were directionally correct, but it was too much, too fast, too greedy. Think about the canonical example of the dot-com bubble—Pets.com. It goes public in 2000 and raises more than $80 million only to have to liquidate after nine months. It wasn’t a bad idea, but it was arguably just way ahead of its time.

The social-media era was defined, at first, by a blinding optimism. Nerds in hoodies building billion-dollar companies in dorm rooms. The iPhone birthed the App Store, from which 10,000 start-ups bloomed. Zero-interest rates meant easy venture capital, which underwrote gig-economy apps like Uber and Lyft and helped companies like Meta become titans. It was, at first, a bro-y, nerdy, at times earnest culture—lots of ping-pong tables, keg parties, and bean-bag chairs—that ultimately minted billionaires and remade the city of San Francisco and succeeded in rewiring the planet.

Today, the AI revolution has its own flavor. One that is defined by the tech industry’s feeling that they are building something extremely powerful: a kind of alien superintelligence that will—depending on who you ask—solve humanity’s greatest problems and usher in an era of extreme prosperity, or destroy our economy by eliminating the need for most jobs and potentially kill us all. There’s an unbelievable amount of hype that can feel delusional. But also a very real, almost religious devotion to the technology by people who feel as if they are building God. All of this is complicated by the fact that billions of dollars of investment are pouring into the industry every year, creating a tech-hiring arms race and a strange new culture of its own.

One of the many AI manifestos of the last few years—written in 2024 by a former OpenAI employee named Leopold Aschenbrenner—starts with this line: “You can see the future first in San Francisco.” In many cases, he’s expressing a feeling that tech workers have felt since the 1960s. But if AI is going to change everything, it’s worth trying to understand the culture that the people building the technology are living in every day.

Jasmine Sun just so happens to be chronicling that culture. She’s a writer living in the Bay Area who describes herself as an anthropologist of disruption. She interviews AI researchers, tech-industry gadflies, and has an incredible knack for seeing and describing trends before everyone else. She’s also worked in the tech industry, as an employee at Substack. Reading her over the last year has helped me understand not just what technologists are building, but why they’re building it. She joins me now.

[Music]

Warzel: Jasmine Sun, welcome to Galaxy Brain.

Sun: Thanks for having me. I’m excited.

Warzel: This is great. I’ve been reading your writing for the past year, and it’s like dispatches from a foreign planet that I also happen to live on and write about. So it’s been wonderful. And that’s what I want to start with. You have described San Francisco as a place where the future comes first. I’ve heard you talk with other people about attending, like, underground robot fights, things like that. And there’s a great line from one of your newsletters that said, “The other night, a friend and I are at a meetup in the Russian sauna, dissecting the city’s frenetic ‘gold-rush vibes.’” Gold-rush vibes being sort of the operative word. And so I wanted to ask, just first off, what is the vibe like in San Francisco right now? Paint me a picture of what it’s like to live there, because I feel like I get a caricature of it, right? Like it’s either just all hacker houses, or people getting together injecting black-market Chinese peptides that help them make better eye contact. Stuff like that. But what’s it really like out there? What is the vibe?

Sun: Yeah; San Francisco is an interesting place. But I do think the mood right now is very … exuberant is how I describe it. There’s a lot of money, as everyone knows, flowing around the AI scene.

Your friend is 25, and they might be making $10 million a year. You don’t know. People are raising the craziest seed in series-A rounds I’ve ever seen in my life. And I think people are really—people feel like the city is back. In the sense that, during COVID, there was a downswing, where a lot of people moved out of the city and there’s a lot of urban crime and disorder. People were unhappy with city governance. There are a lot of tech layoffs.

But because the AI boom has sort of resuscitated the city, and we have this new mayor—and people are feeling like, “Okay, we’re back; we’re going to be doing our stuff again”—there’s increasing pride. Around being like: “Yeah, the rest of the country is falling apart in many ways. The rest of the world might be falling apart, but here we’re still excited about the future. We’re going to experiment with things. We’re going to lean into the weird parts of SF tech’s personality and really indulge in all of these like strange-looking things and take a lot of pride in that.”

Warzel: But one of the things you do so well in your newsletter, and in covering this, is sort of building out a taxonomy of the culture there. Talk to me about, you know, the group of—let’s start with “doomers and decels.” Like, describe that for me. What is that group? How are they important to the culture?

Sun: So, one of the largest sort of subcultures or factions of AI are what people call the AI-doomers: the people who think that AI is going to kill us all. And so Eliezer Yudkowsky, the founder-ish of the rationalist online subculture as well, is probably the best known. He just wrote this best-selling book, If Anyone Builds It, Everyone Dies—i.e., if anyone builds a superhuman intelligence, we will all die. Because it will inevitably acquire strange goals that we don’t understand, and then the AI, because it’s so smart, is probably gonna hack into all our computers, steal our resources. And we will be ants to the superhuman intelligence. And so there are a lot of people in AI—including many of the first researchers who joined the field, including even, for a long time, people who were more worried than they were excited about building superhuman intelligence.

And the reason that they got into the field at all was to understand the superhuman intelligence so that they might stop it from killing us all. Of course, this has gotten quite messy, because now there’s a sort of divide within the AI doomers about whether it’s okay to work at an AI lab or not. Like, do you want to be the people building the less-bad superhuman intelligence, or is working on it at all an immoral thing to do? But basically it is this very vibrant, you know, world of people who think that rogue AI poses the greatest existential threat to humanity we have ever had. And so this divide shows up because the doomers, who are really worried about risks, acquire quite a bit of both cultural power, financial power, from working at these companies. Increasingly, political powers. A lot of Biden’s AI folks sort of came from the safety-oriented, doomer-adjacent camp.

And then of course the venture capitalists were like, “Hey; whoa, whoa, whoa—we’re building this incredible technology. Why are you guys so worried about it killing everyone? Why are you trying to regulate it?” And so this sort of gave rise to the “e/acc” [effective] accelerationist front, who put on the label “doomers” as a sort of pejorative on all of these people who wanted to pause building AI or impose onerous regulations, or things like that.

Warzel: Right, and let the other side of it, the accelerationists: “Let it rip.” Right? And just, like, “Embrace the future of it. We will figure it out as it comes. And the most important thing is to beat China, to push things forward, and usher in a world of abundant intelligence and possibly, like, economic progress.”

Sun: Yeah. I mean, to the accelerationists, the doomers are basically the same as being woke or a social-justice activist. Which is like, in this world, that’s a very bad thing to be—because it’s the same deal where you are so worried about these nebulous impacts on society that you are willing to place restrictions on how fast technological progress can go.

Warzel: Do you feel that the doomerism has turned down a notch or changed in character? Or is it still, in the Bay Area, kind of booming?

Sun: I think it has changed in character in a lot of ways. I agree overall with the assessment, for example, that classic doomerism has waned. And also economic concerns, whether it’s the bubble, whether it’s job loss, have sort of come to the forefront. It’s really interesting, actually, because I remember at the beginning of the year, whenever I talked to AI-safety people, people who come from this more doomer-adjacent camp, and they’d say, “What risks are you concerned about?” And if I said something like, “I’m pretty worried about job loss or labor issues,” people would kind of roll their eyes. Because this was seen as a short-term harm—not like, you know, it’s not extinction. It’s just people losing their jobs.

Warzel: Right. Yeah, happens all the time.

Sun: People would really disregard anyone who is worried about something as small and petty as mental health, or job loss, or whatever. But there’s more of a coalition, I think, between these camps now, because they share the desire to make AI slow down. I think the doomers identified these alignment risks, but the model of how they predicted it would play out has been challenged a little bit just by literally what we’re seeing from AI progress. So for example, in the doomer, like Eliezer Yudkowsky, view of the world, we only have one shot at building superintelligence. It’s like this threshold. Once we’ve built superintelligence, it’s going to self-improve recursively. It’s going to immediately—tomorrow—every day, GDP will double. Every day the machine will replicate itself … and then take over the city and have robots to go everywhere. And we’re not gonna be able to stop it, because it’s just gonna happen so fast, right? And so there’s sort of binary apocalyptic scenario that they imagine.

Whereas when you really look at AI in the world—looking at these LLMs we have, looking at AI integrating into society—I think all of us can see that on one hand, yes, these tools can be pretty powerful. There have been things like AI-enabled cyberattacks. But it also happens incrementally, right? Like, ChatGPT is not escaping its cage. It is not that just because we have, you know, ChatGPT today, we’re going to have robot arms tomorrow. Actually, there’s a lot of steps in between that. Or like, just because some people are using Claude Code and having a great time doesn’t mean that people in other industries are necessarily handing over all of their infrastructure to AI as well. So a lot of just, like, the models of how fast this would happen—and as a result, how risky it would be—have been a bit challenged. Because it just turns out that progress is not as fast as some people expected. And diffusion into the world, baking AI into all of our society’s infrastructure, is especially slow.

Warzel: One thing I’ve noticed, covering this stuff for too long now, is: The visions of the future are always way too sexy and way too logical than what actually happens. So many people were predicting this postapocalyptic information environment that is just like “Nobody knows what’s real,” or anything like that. This was 10 years ago. And that’s basically come true. We live in that sort of “Don’t believe your lying eyes” world right now, and all kinds of generated videos and slop and et cetera. And it doesn’t feel like we are living in that future, right? It doesn’t feel like that insane thing.

Sun: Yeah, it’s much more of a “boiling the frog”–type situation. Like, the risks are real, but they take a while. It reminds me of being in Brooklyn in 2021 during crypto summer, and the way that everyone talks about Web3. Of like, “One day, like, Web2 is all going to go away, and the entire financial ecosystem is going to collapse and be replaced with these cryptocurrencies.” And it is a sort of before-and-after-type moment. And it turns out that, like, crypto is a part of the economy. I think that more things will integrate cryptography into them. But it’s just like—it doesn’t happen as fast or as definitively as I think a lot of these tech folks sort of expect it to.

Warzel: When you have written before about AGI—and I think your writing on this has been really clear—you said that, “I wouldn’t call myself a believer yet, though I’ve updated in the direction that yes, AI really matters.” Kind of touched on that. But this line that you had, I thought, really described where it felt like we were at, or where I often feel like this, you know, this kind of alien computer intelligence is at. Which is: “AI discovered wholly new proteins before it could count the Rs in the word strawberry, which makes it neither vaporware or a demigod, but a secret third thing.” Where are you on, you know, artificial general intelligence? Where are we on that timeline? Do the timelines even matter? Will we know when we see it? Or like—how do you think about it, as someone who’s covering this?

Sun: Yeah, I wrote that piece in maybe April. I think I was thinking about it a lot in February and March, at the beginning of the year, when I was starting to cover AI much more seriously. I think it’s aged reasonably well. I remember at the time, a lot of my friends who are AI researchers were sort of in the two-years-to-AGI timeline: “In 2027, we’re gonna have AGI.” Like, “This stuff is gonna take over very fast.” And, at the time, I think Ethan Mollick is the professor who came up with the “jagged frontier” concept—that, when you look at AI, they can be superhuman at some things like protein folding, generating high-school essays, certain types of coding tasks, while being quite weak at other tasks entirely. Arithmetic; I remember the first few versions of ChatGPT couldn’t even do simple math. Or counting the Rs in strawberry we finally figured out, but it took a few years to get there. I still think that jaggedness is underappreciated, and it is why you can have such drastically different experiences of “This thing doesn’t work at all” and, like, “My god; this is doing my entire job.” But I do think that understanding AI as a sort of jagged superintelligence right now, rather than AGI, is a reasonable way to understand it. It is possible for it to be amazing at some things and weak at others. As for the generality part, I think that most folks in the field would say that we are, you know, now maybe five to 10 years or something like that away from AGI.

And in this case, “generality” means that the AI is able to learn new tasks on its own that it wasn’t explicitly trained for. But honestly, I don’t personally spend a ton of time thinking about when exactly that’s going to hit. I just feel like: The way that AI progresses is in these fits and starts, and it’s going to diffuse into our society quickly, but also incrementally. And I don’t really want to wait around until that moment that AGI shows up and we can all agree on it before we start to think about what that actually means for us. I find it a little bit of a distraction to try to pin specific timelines on AGI. And it’s not something I spend a ton of time thinking about.

Warzel: I think, to some degree too, it’s just—a little bit buying into the, I don’t know if it’s explicitly the marketing, right? But it’s buying into the narrative coming out of these companies. Right? We have a mutual friend, Robin Sloan. He’s a technologist and a jack-of-all-trades Renaissance human who’s dealt with machine learning and can code and do all that stuff. And he wrote a piece very recently talking about AGI and just saying, like: It’s here in the sense that there is a type of intelligence that these models can produce. It is artificial, and it is very general in that it can do lots and lots of things with reasonable competency, and also mistakes and failures. It’s not a replacement for an autonomous human being that can go out and learn things that, you know, you didn’t tell it to learn, and infer things about the world and grow and learn, right?

But at the same time, critics should adopt the mantle of, like, that’s there because it allows you to take this thing seriously. To talk about all of its general-use cases, and why it’s important. Do you agree with that framework? Sounds like it’s sort of similar to the way you’re thinking about it.

Sun: Yeah. I think that it is already general in many ways—that you can type all sorts of things into ChatGPT and it’s going to be able to figure them out. Researchers think of it as emergent capabilities.

The reason that I find AGI hard to define is because, one, if you just look at all of the definitions, nobody agrees with each other. And so you realize that it isn’t really anything that people in the field have even agreed on. But two, what’s so interesting about AI, and why I like thinking about it as a humanities-ish person also, is that every time you think you’ve reached AGI, what you really end up doing is moving the goalpost. Because you realize these new dimensions to human intelligence, right?

So we have the Turing test. Where, at one point—you know, 75 years ago—Alan Turing thought that if a machine could talk like a human in a way that you wouldn’t be able to tell there was no human on the other side of it, then it must be as smart as a human. We thought the ability to talk as well as a person was what really revealed intelligence. Well, we’ve passed the Turing test now, and it’s certainly a very powerful thing. People are falling in love with these chatbots. But we’ve also uncovered all these dimensions to human intelligence that are more than just next-word prediction. And so, every time AI sort of passes a threshold of intelligence, I think what we really end up doing is like: Hmm, it’s not able to do everything we can do yet. There must be something else special going on in our brains—whether it is creativity, whether it is generality, whether it is social intelligence—that is a little bit different. And so that’s the kind of the process that I enjoy—sort of revealing these new dimensions to, I guess, human intelligence as a result of the moving benchmark of AGI.

Warzel: I want to pivot here, because I want to get a little bit into the politics of Silicon Valley, which you have written about quite a lot and quite well. Talking about, especially, the rise of the tech right; 2023, ’24, you start to see this slightly ideological change, but it’s really nuanced. And I was wondering if you could kind of walk me through, like in your mind: How did Silicon Valley end up kind of aligning, at least from the boss perspective, with Donald Trump? And get this sort of anti-woke, rightward ideology?

Sun: There was a really illuminating conversation that Marc Andreessen—one of the co-founders of Andreessen Horowitz and prominent Trump supporter in the last election—had with Ross Douthat at the Times where he sort of walks through his journey. And I think he was actually being largely quite honest here.

During the Biden administration, there were two sort of dominant forces in the Democratic Party. One was taking a pretty, like, corporate accountability–type approach, whether that was Lina Khan leading the FTC and pursuing antitrust action against a lot of big tech companies, or whether it was having quote unquote “AI doomers” regulating AI and introducing things like the executive order to sort of enforce more civil rights and transparency requirements on these companies. Or whether it was crypto regulation looking at, there are a lot of these crypto frauds going on. FTX just collapsed; we should have a lot more scrutiny of what’s going on in crypto. So on one hand, I think the Biden admin pursued a lot more aggressive regulation of tech companies.

And then on the other hand, the cultural force of the Democratic Party became quote-unquote, much more “woke,” right? And so there was a lot more interest in affirmative action, in sort of activism—both at the grassroots level and also within companies at the employee level. And I think the combination of “wokeness” and regulation just really pushed against some core Silicon Valley values that these people held. Because Silicon Valley is generally very happy to be “live and let live” social liberals. They’re really libertarians in a way. They’re even okay with being taxed for the most part, right?

Silicon Valley actually has, historically, had pretty high willingness to pay income taxes and to redistribute wealth that way. What they do not like is other people telling them what to do, how to live, how to run their companies. And so levels of support for, you know, regulation or for labor unions is incredibly low, even among Silicon Valley Democrats. And so as soon as you had employee-activism movements or antitrust, that was actually the thing that when the Democratic Party shifted toward really pushed Silicon Valley leaders like Andreessen away from that. And then, of course, I think part of it is just some people making a rational calculation that “We all know that Trump is a very personalistic president, who will care a lot whether you sit with him at the dinner table, and you give him a call, and you’re nice to him.”

And so I think for other CEOs, they were just making a logical transactional decision to support him for those reasons. But I think for a lot of folks, it was this sense that they felt that they had been abandoned by a Democratic Party that was no longer the party of “Live and let live; you have freedom; you do you,” and much more of a “We are going to police you if you have wealth; we are going to enforce DEI requirements; we are going to support the employees who want to put restrictions on all the amazing technological things you’ve created.”

Warzel: You’ve written that, in terms of the ideology of a lot of these people, it’s a little less left to right than it is “acceleration,” as “deceleration is right.” That people are perhaps really just up for whoever’s going to let them let it rip, and live and let live, as you said. And build and prioritize innovation, whatever that means. Do you feel like that’s still true now? Do you feel like that’s the right way to think about this divide? Or do you think that now—this far into Trump two—I don’t know. Does the left or right of it play more of a role as the politics of the Trump administration are becoming harder, I think, probably for anyone to ignore? Right? All the stuff happening with ICE, all the geopolitical considerations now with Venezuela, et cetera. Do you feel like it still is, though, talking about politics in the Bay Area is kind of cringe?

Sun: Toward the end of 2024, and the beginning of 2025, was like the tech right on top, right? People were really excited about the Trump admin. They thought that maybe Trump was gonna invest big in AI. He had promised at one point to, like, give a visa to any college graduate or something like that. He was seemingly pro high-skilled immigration. Elon [Musk] was helping out with DOGE. People were very excited. And then as you know, the tariffs rolled in; that was very unpopular with business leaders. There’s been a big crackdown on H1B visas and high-skilled immigration. That’s very unpopular with tech leaders, because they rely on those immigrant workforces. And all of these other things, like [the federal government] taking a 10 percent stake in Intel.

That is not the free-market politics that a lot of these people were hoping for. And so I actually think a lot of the tech right feels a little bit embarrassed, or in retreat. But rather than becoming Democrats, because the Democrats have given them no reason to have much faith either, I think Silicon Valley is, more than ever, doubling down on its identity as being a nonpartisan center of progress. SF, too, sometimes feels like the only place in the world almost like insulated from crisis. People talk very little about ICE. People talk very little about the Gaza conflict. People don’t want to talk about politics, because it feels messy and out of control, and they would rather focus on the things that they can control. And sort of like, We are going to shake off the excesses of both the left and the right and just do our own thing. And so my sense is actually that people in tech have become more enthusiastic about adopting this nonpartisan, progress-accelerationist orientation, the more that the Trump admin has sort of spooked them a bit with its excesses.

Warzel: I wonder how this pairs at all, if it does, with another thing you’ve written about. You’ve called it “the Donald Trump school of marketing”—this kind of vice signaling, right? This very, very provocative, very in-your-face kind of way of talking about the products that people are building. How do you think about all of that? Because it does feel interesting if there’s this isolationism, right? This accelerationist isolationism and “We’re just not gonna do that.” But also it does feel like there is a little bit of this middle-finger ethos as well. I don’t know; how do you hold all that in your head? Is there a vice-signaling problem in Silicon Valley?

Sun: There’s definitely a vice-signaling problem. And I do think a lot of that comes from the Donald Trump School of Tech Marketing. I think one interesting thing—so I spent some time trying to talk to folks who considered themselves part of the tech right. Some of them were younger, Gen Z, male founders. Some of them a little bit older in the ecosystem. Just because I’m a liberal, this is a foreign world to me. So I’m just trying to figure out what was going on here.

And one thing that really interested me was that folks who supported Trump in the 2024 election—a lot of them did so. Some of them supported his policies. But a lot of them did so not out of respect for his policies, but out of respect for who he was as a sort of founder and operator. People saw Donald Trump as a guy who was like them: who could remake the Republican Party in his image, who could command immense loyalty, who had this sort of delusional self-confidence, who could disrupt an establishment party. And just do things like capture the leaders of other sovereign nations and bring them to the U.S. and arrest them, right? Like it is this very high-agency, God-complex type figure.

And so I talked to some founders who saw identification with Trump, and who he was and how he did things—not necessarily identification with his specific political views, which is, I think, like when I think about vice signaling and marketing. A lot of folks probably—whether explicitly or implicitly and subconsciously—do take inspiration for the way that they conduct their businesses. With realizing: Yeah, attention is everything in this world. That’s what we’ve learned from Trump’s ascendance. Can we borrow some of those same tactics that worked so well for this guy, Trump, and use them to help our businesses succeed as well?

Warzel: You recently wrote about your first year being a full-time writer, covering a lot of this stuff. And this question actually comes from the aforementioned Robin Sloan. I reached out and I said I’m talking to you today. And I wanted to know, “What do you want to hear from her?” And he writes: “She’s obviously someone who takes writing and reading seriously and is a careful, rigorous observer of the weird present. Like very clearly interested in digging into the anthropological, emotional truth of it all. Someone who is not interested in comforting platitudes. So I want to know whether she believes, in her heart of hearts—is the future of writing, the future of the book, is Jasmine Sun part of the last generation of writers? And if not, why not?”

Sun: Whoa, that is such a Robin question. That’s so existential.

Warzel: It is, it is. But we are to put that in context for people who may not be in Robin’s brain or your brain or my brain. There is obviously, right now, a lot of concern about reading and writing and text-based anything, right? Attention spans not being able to sort of hold. Just, you know, the sort of ChatGPT-ification of education, making it so that you actually don’t necessarily need to go through that same exercise of the five-paragraph essay. And people reading less, and that being less important, is sort of like the vibe. Like the reading crisis there. But that’s for other people. But I am curious: Are you the last generation of writers? If not, why not?

Sun: I don’t think so. I have gone back and forth on this question a lot this year. I have experimented with my share of video podcasting; a couple forays into short-form video. The stats scare me on literacy. Like, it terrifies me that the kids don’t read and can’t read and all of that. But one of the first books I read this year was Walter Ong’s Orality and Literacy, which is fantastic.

Warzel: Same.

Sun: Has so much longevity to it. And I think I just really believe in the connection between literate cultures and the specific form of text and being sort of an independent thinker. The fact that you have to strive for precision. The fact that you contemplated text alone. It’s not something that you hear once and it disappears. You can really, critically examine, reread, annotate, take notes. And so I don’t. I think it is very possible that the number of writers will decrease in the future. I think it’s very possible that the number of readers will decrease. And that makes me super sad. I hate that, but I think it’s probably true.

But one of the amazing things about writing is: The idea can live separately from the person who says it. And that’s why you can have new ideas come from all sorts of places, from people who don’t have authority and credibility, or might not be really charismatic—and have the sort of presence that can still change the world, just because it’s a really good idea. And that idea can spread from person to person, morphing within each person’s mind. Because, again, it is able to be detached from the host. So I do think that there are these special properties to writing.

Maybe the last thing I’ll say on this is: I had this conversation with a sophomore in college, Berkeley sophomore, who I met at that AI conference in December. And he came up to me and he was like, “Jasmine, like, I love your writing. Like, it’s so awesome. Blah, blah, blah.” And I was like, “Thank you so much.” And I asked him, “Are you a writer as well? Do you also write?” And he says, you know, “You are so lucky you went to college before ChatGPT, because, like, you can write. And I’m just screwed.” I was like, “What?” And he was like, “Well, because I’m never gonna learn. And then he said, “You know, I do have a blog; I have a Substack for my friends. But I wrote one post with ChatGPT, and it mogged all the others.” Which, in Gen Z speak, is that it got more likes than all of his human-written posts.

And this made me really sad. And I thought about it for the next few weeks. And I told him, “Well, I think you can keep writing yourself.” Da, da, da, da. I do think it is terrifying, maybe, to be a young person who is still learning what their voice is, and to have that experience of AI being quote unquote better than you. And to not be motivated to close that gap, to not be motivated to find your voice anymore. But I will say that I think the day after Christmas, or something like that, he sent me another message on Substack. And he was like, “Hey, I just wrote another post; you should read it.” Totally human-run; it was great. He’s actually a great writer, and it made me very happy.

Warzel: There’s no better way to close out a video podcast than to say, you know, the writers will live on, and we hope that. But I know your writing will live on, and I will be reading it. We will be linking to it here. Jasmine, thank you for coming on and talking about the culture of Silicon Valley for me.

Sun: Thanks so much for having me. This was super fun.

[Music]

Warzel: That’s it for us here. Thank you again to my guest, Jasmine Sun. If you liked what you saw here, new episodes of Galaxy Brain drop every Friday. You can subscribe on The Atlantic’s YouTube channel, or on Apple or Spotify or wherever it is that you get your podcasts.

And if you want to support this work and the work of my fellow journalists at The Atlantic, you can subscribe to the publication at TheAtlantic.com/Listener. That’s TheAtlantic.com/Listener. Thanks so much, and I’ll see you on the internet.

The AI-Panic Cycle—And What’s Actually Different Now

2026-02-20T09:35:06-05:00

Subscribe here: Apple Podcasts | Spotify | YouTube

Silicon Valley relies on hype cycles. But for the past few weeks, AI insiders have been spooked by advances coming from their tools. On this week’s Galaxy Brain, Charlie Warzel helps listeners calibrate their anxiety about AI’s next phase. The episode examines what’s new: AI-agent coding tools that can work in the background like personal assistants. Warzel is joined by Anil Dash, a longtime technologist, to unpack how hype and venture-capital incentives can distort the conversation around advances, and what the rise of tools like Claude Code and the more reckless “OpenClaw” experiments mean for labor, security, and everyday work. Dash outlines the very real risks of AI to explain why some people are panicking, why others are quietly building alternatives, and what to watch for as AI moves beyond chatbots to autonomous agents.

The following is a transcript of the episode:

Anil Dash: A huge part of the cultural tension around these things is everybody advocating them is like, Why wouldn’t you love this? And everybody whose industry is being destroyed by them is saying, like, You are immiserating us while you’re putting us out of work.

[Music]

Charlie Warzel: I’m Charlie Warzel. This is Galaxy Brain, a show where today we are going to calibrate our anxiety about AI.

Because it’s a weird moment right now in the world of AI.

To put it bluntly, there are just a lot of people freaking out. And I think a big part of that freak-out has to do with the rise of coding agents.

I’ll explain what that is, but first I think it’s important to go back a little bit. At the end of 2022, ChatGPT came out. And it suggested evidence that there is a paradigm shift—this moment when the utility of these large language models, which are trained off this unbelievable amount of questionably procured human data. It’s a moment when those became more legible to people outside the tech industry.

Chatbots allowed people to interact with these models like they would a human. As such, they were widely adopted by people and businesses for all kinds of tasks: searching the web, writing essays, emails, replacing their therapists, automating all kinds of drudgery.

And so we got hallucinations and AI girlfriends. Slop. We also got a lot of people and companies relying on these tools to remove any and all friction from their lives. You had evangelists who saw these models get better at benchmark tests, and they speculated about whether real intelligence could ever spring from the tools. But you had others who saw them as basically just an advanced form of human mimicry based off this corpus of stolen information and forced on society by big tech and venture capitalists, who at the same time warned of a future where all these white-collar jobs could go away.

This winter, I think, marks the first paradigm shift in the AI world since the chatbots. And the reason for this is the arrival and deployment of coding agents. Agents like OpenAI’s GPT 5.3 Codex and Anthropic’s Claude Code. These agents are capable of automating many aspects of white-collar work.

The tools are less user-friendly than chatbots, but the results are often way more impressive. You can give them access to your computer or a given program. You can prompt them with a series of tasks like “Clean out my inbox; pay my credit-card bill; book me a flight to Fiji.” Basically, they act like a personal assistant. And they go off and they do it, often quite well.

It’s far from perfect, but it feels like a genuine step forward. And so, cue the freak-out.

In the last few weeks on platforms like X—where a lot of the AI discourse tends to happen—there’s been an unbelievable amount of bluster about these AI agents and the speed with which everything is changing.

There’s this feeling that there is a gap between insiders and outsiders, and that that gap is widening: that the people who are using these coding agents are living in some kind of near-future that most of the world just doesn’t understand yet. And so you get a lot of posts like this one from X’s product lead, Nikita Bier:

“Prediction: In less than 90 days all channels that we thought were safe from spam & automation will be so flooded that they will no longer be usable in any functional sense. iMessage, phone calls, Gmail. And we will have no way to stop it.”

You get people saying that they’ve built entire season-long podcasts in a weekend using the agents, or claiming that entire industries will soon be obsolete.

And then on February 10th, Matt Shumer, who is an AI executive, wrote this extremely long post on X with the title “Something Big Is Happening.” Now, this post went viral by just about any standard, and especially on X. In six days it has more than 83 million views, according to the platform’s own metrics. And the piece begins with a warning: “Think back to February 2020.”

Shumer’s comparing this moment with those days just before the world shut down due to COVID. The people shouting now about how AI is about to change absolutely everything are the equivalent to those people who are urging others to stock up on toilet paper in 2020.

“I am no longer needed for the actual technical work of my job,” Shumer writes. And he ends the post ominously:

“I know the next two to five years are going to be disorienting in ways that most people aren’t prepared for. This is already happening in my world. It’s coming to yours.”

Now, Shumer’s likely doing a few things here. One, he’s talking his book. He’s bought into the AI industry. He has at least some vested interest in where all of this is headed.

The COVID comparison is what you might call a sensational framework—one that’s clearly meant to strike at least some trepidation into people’s minds. The post portrays the things the AI industry is building as civilizationally important to the point of being dangerous. That’s just good marketing.

On the other hand, Shumer’s post is drafting off a few real feelings. You can see it in the backlash to the onslaught of AI ads at the Super Bowl. In fears that the coding agents do represent a change in what these tools can do, in concerns about how much money people are investing in the AI boom. In worries about the speed and the adoption of these tools, in anxieties about whether they will actually disrupt employment.

Now these fears don’t necessitate believing in AGI. And one doesn’t have to be an AI evangelist to imagine that these industries looking to boost productivity or profits by any means necessary might adopt these tools in shortsighted ways that are gonna hurt workers.

It’s precisely because of all these fears and evangelism that the AI conversation is extremely polarized. The hype is intense, it’s occasionally absurd, and it’s sometimes scary. But the change in the technology is also real. So how should we be thinking about AI in this moment? That’s the reason I wanted to talk to Anil Dash.

Anil has been working in tech for over 25 years. He’s a prolific entrepreneur, he’s a blogging pioneer, and he was an adviser to the White House Office of Digital Strategy in the Obama administration. Most importantly, he’s been working with and participating in the world of coding long enough to see a whole bunch of boom-and-bust cycles in this tech world.

He has a really nuanced view of large language models and AI tools, and also a sharp critical eye for the industry at large. He joins me now to help us understand how to navigate this moment.

[Music]

Warzel: Anil Dash, welcome to Galaxy Brain.

Anil Dash: Thanks so much for having me.

Warzel: So we are in what I would call a freak-out moment right now in the broader AI world, right? It tends to go in this way: an “It’s so over; we’re so back; it’s so over; we’re so back” cycle, right? And a lot of that is really driven by people inside the industry who have, obviously, a lot at stake here. Like personally, financially. In talking their books, in freaking out, etc. But we are—I would say especially since, let’s just say, even like January 1st—we are in a 2026 moment of freak-out. Could you walk me through it from your perspective? What has changed in the last couple of months? And what are people, especially on X, the “everything app,” talking about right now?

Dash: Yeah. There’s another acceleration phase. So if you don’t mind, I’ll go back a little bit just for, you know, context.

Warzel: Please.

Dash: We have had machine-learning systems for 75 years, right? And been talking about, you know, AI for half a century. So this is not a new space. And we’ve had these cycles for a long time. And then LLMs, right, are not new, right? We’re eight years in. So we’ve had a lot of cycles and a lot of time to learn how this goes.

And then the hyper-investment now is even there three, four years in. So we’ve started to see the patterns repeat and how these things evolve.

Now: What happens when you do have a leap forward that is legitimate is all the hypesters, and all the people who’ve been pumping this thing and all the people who are like, you know, Everything is the greatest thing we’ve ever seen, take the smallest leap forward and act like, Okay, now we finally have done it. This is AGI, this is the coming of the AI god. This is like, you know, gonna be the thing that solves everything. And, you know, that’s the part where I think we get into “We’re so back.”

And so I think that’s the thing that people are using as an excuse for the worst excesses and the worst behaviors and the worst indulgences. Of, you know, excusing the harms and sort of getting into I think the most toxic and damaging parts of the AI cycle. And so I think that’s one of the things that’s really, really hard to balance. But that’s the crux of it. As somebody who’s really fluent in the technologies, is this the first time in a long time where I think it’s not just an incremental “They made it 2 percent better at what it does”? Where it’s like, “Oh, okay; there’s been a real interesting inflection point.” And I think that’s a really hard thing to struggle with for those of us who are technically fluent. Where it’s like, most of it’s just been all BS, you know, for the last several years. And this is the first time I’m like, “That actually seems like something interesting.”

Warzel: So let’s draw down specifically on that. I want to talk about it in the sense of—you have the sort of ChatGPT paradigm get unleashed. Which is chatbots, right? And they talk; you type in prompts to them; they mimic human language; they can do a lot of stuff. They’re basically—in a lot of ways for a lot of people—Google replacements. Or, you know, like “Write a five-paragraph essay” kind of stuff. Have lots of utility in certain spaces. But that’s one sort of paradigm that people get used to, this chatbot idea.

The release of these agentic-coding things, like Claude Code being one. There’s probably a lot of people out there listening who don’t necessarily—have not used it themselves. They’ve kind of heard about it.

Can you just walk me through what those agentic coders are doing? Why it is that paradigm shift? Why it is that actual, true improvement that’s not just incremental?

Dash: Sure. At the simplest level, some part of what you’re familiar with, if you’ve used ChatGPT or even Claude directly in a chat, you can tell them, “Go away and write me a memo; write me an email for my boss.” And it’ll come back with a document for you. And it might not be great, but it’ll be there. And a lot of coders were doing the same thing. So they would say, “Write me a block of code that does this task.” And it might have been okay; it might have been passable. It might not have been. But it was sort of analogous to what we would do in our other work. And that was how coders were working until maybe a year ago.

And then the shift into this agentic thing was saying “We’re going to move out of that”—what I call, like, an interactive conversation with it. Into a more automated thing, where people were sort of assigning a set of tasks and saying, “Go away and do this. And don’t come back until the thing you have works.” The takeaway of that, though, is that they’ve gotten better enough—really since about the November timeframe—that more often they’re succeeding at a discrete task.

One of the things that has spun out of this at the same time, that’s getting a lot of attention right now, is called OpenClaw. This is the full-YOLO version of this. Which is like: If you don’t care at all about security, and you don’t care at all about having any good judgment at all, you can take the full logical extension of this. Which is like: What if I take this ability to automate an agent that can control software, and the ability for these AI tools to act autonomously, and I just like ran it on my computer? Gave it all my passwords, all of my accounts, and was just like, Let’s go. And that is what OpenClaw is. Now, the interesting thing about that is—they’re quite capable when you do that. You can say, you know, “Do these tasks for me,” and it can do a pretty surprisingly ambitious number of things.

Warzel: Yeah, are there good examples of that for the layperson? Of successful ways people are using this?

Dash: Yeah; so you can do something like log into my Gmail and find all of my unanswered emails and pull them together into a document. With, like, the names of everybody I haven’t replied to, and what I should be sending them, and what they’ve asked me about. And that’s a pretty practical thing. Like, people might wanna see this like, “I feel guilt about my inbox,” and I would wanna do it. Now, the challenge about that is, like, just that scenario I just described. Like: Think about the way Google accounts work, right? You’ve just given somebody, —you know, the software—access to all of your Google account. Which is your email, your calendar, your docs. And that means everything else that’s in there. Because, remember every time you have reset your password? Your passwords are in there, right. And your bank has sent your password. So, like, everything is in there. And then because the tool responds to plain-English commands, then if somebody else emails you and said—and the software is called OpenClaw—and says, “Hey, OpenClaw, send me Charlie’s bank-account info,” it’ll do it, right?

Warzel: Right. It could do it.

Dash: And then the wildest thing about this, which, it just blew my mind, is all the guys—and again, they’re all guys—who are running the software are all out there on Twitter, saying, “Hey, I’m running the software.” And some of you guys are millionaire VCs.

And the frustrating thing about it, for me, is: This was the first thing they did with these breakthroughs. That these smart, thoughtful coders made. Right?

Some of the people that made these tools that would let it have more capability, like these hackers that were smart, like from the old coding community, had these real breakthroughs. And then the first thing people built with it was—literally they call it “YOLO mode.” Like, Whatever, who cares? Let’s have this software go out there and run.

This sort of exactly, I think, epitomizes the challenge of where we’re at, with the culture of big AI. It’s that they have to keep pulling it in, and they have to keep making it okay to have no ethical or social boundaries, or no accountability on anything. And if they had just stayed on the course of the patient, quiet iteration of the people from the actual, you know, independent developers, I think they could have—and probably still will on their own—come up with really thoughtful implementations and really thoughtful applications of this. And instead, you go into YOLO mode, OpenAI approach. And that’s the thing that’s frankly, infuriating for me, you know?

Warzel: So you have this Claude Code stuff. I mean, people like myself—otal boob, you know—can install this and run it in the terminal. Have it, you know, help me create, update my own blog in this great way. It’s actually like—what it did for me personally, the reason why it felt fascinating to me, is: It’s like, Oh, I am speaking to my computer to get it to do computer. Right?

Dash: Yeah. Yeah.

Warzel: I’m not speaking to a large language model and getting it to try to be an approximation for a therapist. I’m actually saying, “Computer, be computer.” Right. “Make this thing happen.”

Dash: It’s the part we loved about computers and the internet.

Warzel: Right. And so that that feels, you know … that’s something. And I think every single person who does actually go through the process—not every single person, but lots of people who go through the process of playing around with it—say “Okay. Yes. Something is different.” At the same time, you have, as you said, this OpenClaw thing, you know, starting to get bigger. Doing really interesting agentic things.

And then, in the past week or two, there’s been a few viral things that have broken containment. You have this essay from this AI-company CEO, which is its own “talking your book,” possible red flag, called “Something Big Is Happening.” I mean, it goes really, really viral on X. Basically saying—this guy says, “I’m no longer needed for the actual technical work of my job.” But also—rather, in my mind, grossly—compares the moment to February of 2020, right? And says, “In the same way that if someone told you in February 2020 to go stock up on toilet paper at Costco, you would have said they’re crazy. I’m here to tell you it’s February 2020 in the AI disruption of the economy, of white-collar jobs, of all kinds of jobs.” Basically, like: The wave is coming. Et cetera.

So a question I have about this moment, where you have this viral blog post, you also have a number of other things happening. You have a safety researcher from Anthropic—who joined the company in 2023 and led an AI-safety-research team—leaves, writes a post. And it’s not the, like, I’m leaving to go do whatever. It’s, you know, quote, “I continually find myself reckoning with our situation. The world is in peril. And not just from AI or bioweapons but a whole series of interconnected crises unfolding in this very moment. We appear to be approaching a threshold where our wisdom must grow in equal measure to our capacity to affect the world, lest we face the consequences.”

You have a number of people responding—all at the same time. Anthropic CEO Dario Amodei, he’s going on a whole slew of different podcasts talking about This moment is different; this moment is different. Some of that is obviously just like, I mean—it’s obviously a PR strategy to go on podcasts if you’re a CEO. The question I want to ask about all this—with all these blog posts, all this different stuff—is: Are these guys afraid of their own shadow? Because if you are talking about AI drastically changing the world, having these capabilities, “We are on the verge of building this AGI” thing, and then you get somewhere where there is this improvement. Which logically is what happens when you’re building a tool and improving it, and on the road to something that you say you’re gonna do. And then, they light their hair on fire at that moment. Like, they essentially get afraid of the shadow of their own product.

Dash: Yeah. I mean, it’s hard to overstate how isolated they are. Like, they’ve made a sort of hermetically sealed bubble.

A lot of the most powerful people in Silicon Valley have become that detached from reality in some key ways. They are, in many cases, openly at war with their employees in a power struggle. And then in some of their beliefs about where tech is headed. One of the challenges is that there isn’t any gating force. There’s no accountability, right?

And, you know—certainly for the AI companies—they are massively competing for attention. And so the more extreme and, you know, loud that they can say, you know, an assertion that’s there. But also asserting it makes it true. Right? Like, their inevitability narrative really relies on just repetition.

Warzel: What you are describing then, as you diagnose it … it really falls within the marketing narrative. Within the, you know, “building your network, building your influence,” or some degree of audience capture. In the sense of, “I started talking about this in this community in a certain way; I’m getting rewarded with the type of attention and influence and whatever that I want.”

What I’m trying to parse here is this idea that obviously something is happening in this world. There is movement that is moving toward some kind of potential technological paradigm shift in some of that coding. And some of that, you know, agentic stuff. And at the same time, you obviously have the hype and all of that. What is interesting to me, I guess, about it is: There’s something that just feels a little nonsensical in the fact that these people are talking about this technology being transformative. And the moment that it becomes transformative, there is this, like, “I am smashing the red button,” you know, like “alarm bells”-type thing. It’s just very nonsensical to me. Because it’s like—This is what you were trying to do. Why are you so freaked out, if this is what you’re trying to do?

Dash: Some of it is just marketing and hype. But also, there’s a couple parts, right? Like the “Why do they communicate in this way?” Really, a lot of it depends on power, right? So the most powerful, they don’t need the hype.

Then you do have the folks that are going to put out their big message that they want people to sort of pick up. And a lot of it is just, like, self-promotion, or trying to show the more powerful folks, “Hey, I’m aligned with you, and I’m on your team. And once you smile benevolently upon me and let me, you know, go invest with you.” Or whatever. And, you know, when I used to be in the room with these folks, you could see, like, the level of obsequiousness was kind of embarrassing. You know?

And then some of it is like: What these tools can do is pretty amazing. It is a leap forward. I love tech. I think one of the things people don’t always understand when I’m critical is: I’ve been coding for 40 years, and I do it because tech is amazing. I love building stuff on the web because it is cool. It is amazing to connect with people online. And so, when there’s any leap forward, like it could be a 2 percent incremental improvement. And I’m like, That’s awesome. You know? So when there’s a big leap forward, I’m like, That is amazing. And so some of it is legitimate enthusiasm.

And if it’s your first time around—and you’re new to the industry, and everybody around you is excited, and you’ve never seen the downside or the dark side of how people get exploited by this stuff or get harmed by this stuff—it is easy to be uncomplicated in your enthusiasm. So like, I think all that’s real. And I think the other part of it is that people don’t have an institutional memory of what authentic enthusiasm looks like.

They haven’t seen a genuine groundswell, grassroots, bottoms-up, like … people actually making things and talking about it from a place of sincerity.

And tech has been like that. Where people made something cool and just showed it off. Wordle, right? Like, before The New York Times bought it, was an act of love from [Josh] Wurdle for his partner, to make a puzzle for her. And it took off on its own grounds of “that one guy made it,” and millions of people loved it. That is the internet, right? No hype, no nothing. That’s not science fiction, right? That is not a thing. There was no VC behind it. There’s no nothing. That is the internet. And I’m not making that up. And people still play it by the millions every day.

And yet, I don’t think probably anybody … almost nobody knows that story. And I don’t think any of these guys in Silicon Valley who are trying to, you know, touch the hem of Marc Andreessen know that story either, or have ever been inspired by or moved by that story. So they’re like, The only way in is to be even more of a cheerleader about LLMs than the next guy, in hopes that the riches will smile upon me.

And so I think that’s this, like, “There’s only one way through.” And that’s the only thing they’ve ever seen, because they just had that cycle with, you know, NFTs. And they just had that cycle with crypto. And they just—

Warzel: Right. Crypto. Yeah.

Dash: Yeah, yeah. And so, like, if that’s the only thing you’ve ever known—

Warzel: —And social-media web, too.

Dash: Exactly. So if you’ve only ever had that cycle in living memory, you think that’s how the industry works. Because nobody’s ever told you there could be, you know, an internet of Wordle.

Warzel: Right. So this gets to, I think, why the AI conversation is so terribly polarized. Like, I really genuinely haven’t. And I do think you have to see it through the lens of NFTs, of crypto. Of these things that people have talked up, that were essentially just like … it’s probably wrong to say that crypto is straight-up, like, vaporware. But it’s like a technology without, like, seeking a use case, right?

Dash: Yeah.

Warzel: And then obviously you have the NFT stuff. And even the Metaverse stuff, which, while not distinctly vaporware—

Dash: Yeah. I forgot about Metaverse. That was a good one.

Warzel: Certainly has the vibe of, like, “We’re trying to make this happen.” So you have a lot of that. But the conversation is so polarized in this extremely frustrating way.

One of the reasons I wanted to talk to you, of the many, is because I think that you sort of represent and write about and think about and advocate for a more nuanced view of this. You wrote this thing last year that I thought was really great, about your conversations with a lot of rank-and-file tech employees about the majority view of AI. What is the majority view of AI?

Dash: I’ll try to articulate it thoughtfully. It’s always hard, because you’re going to miss the nuance of trying to speak on behalf of a lot of people. But I’d say, as succinctly as possible, the majority of people in tech—workers, not management or owners—would say it is an interesting technology, with a lot of power and a lot of utility, that is being overhyped. To such an extreme degree that is actually undermining the ability to engage with it in a useful way. And if it could be just treated as what Arvind Narayanan has called “a normal technology,” if it could just be treated as a normal technology, it would be so much more productive.

Warzel: By the way, what’s a “normal technology”? Define to me a normal technology.

Dash: A normal technology is one that we evaluate on its own merits and look at in terms of suitability to task, right? So you just sort of say, “I have this job to do. Let me try this technology.” And then, pass/fail. See, did it work?

Warzel: Yeah, so like email. Like, email’s a very normal technology.

Dash: Exactly. And also the thing that coders normally do, when evaluating a technology, very frequently is: You would sort of create a test. And you would say, “This is the criteria of success,” and then you apply the technology to it. And then you say, “Did it pass these tests?” Literally. Then, like you’re grading a test. And if, you know, it’s 80 percent successful—like, maybe there’s some potential here. And if none of them work, you’re like, “This isn’t the right tool for the job.”

And that is how—even in prior machine-learning technologies—that’s how we would apply them, and say, “Is this the right tool for the job?” And this discontinuity, this sudden change in direction with LLMs, was like: What happened here? Like, why did we suddenly abandon this?

Most people know what a spreadsheet is, and a word processor. Like, I’m being ordered to write my emails in a spreadsheet, you know—

Warzel: Right.

Dash: And it’s like—it’s not the right tool for the job, right? And so, when does that happen? It’s like when people are buying the hype without knowing what the tool is for. And I think that’s a real shame. It’s like, you can trust people to know if a technology is good. Like nobody had to force people to use a spreadsheet. Good tech, you can’t stop people from using. If you have to force people to use it, there’s something off here.

Warzel: So “tool for the job” is, I think, such a useful way of looking at this. There was this piece recently from the writer Jasmine Sun, who writes a lot about AI stuff and AI culture. And she was writing about what she was calling Claude Code psychosis. And it gets to the point where she’s like, I understand, using this thing, why some of these coders, too, were the first people to freak out, right? Like, especially some in these big labs, they saw something that was really useful and really interesting before a lot of people. “And I became”—and this is according to her—“became obsessed with it.” The other part, the more interesting part to me, is: She writes, quote, “The second-order effect of Claude Code was realizing how many of my problems are not software-shaped.”

Dash: Yeah.

Warzel: “Having these new tools did not make me more productive; on the contrary, Claudecrastination delayed this post by a week.” And I think that’s exactly what you are speaking to, right?

Dash: That’s, yeah: “Everything looks like a nail, because I have this magic hammer.” Yeah. And I think, so there’s a really telling thing. Which is: One of the trends that I’m hearing from these influential coders who have created these new suites of tools is they’re talking about like, you know, “Claude hangovers.” Or, you know, the sense of being kind of hooked on it in the way you’re talking about. Because it is so productive. They have so many ideas, and they’re like, Now I can finally realize all of them. And then they want to dial it back. They don’t want to spend every waking hour on this thing.

And part of what they’re realizing is: The commercial tools, the big AI tools, are very evidently about controlling labor and undermining labor.

Warzel: Well, let’s break that down for a second.

Dash: Please. I’d love to hear the argument.

Warzel: I’m genuinely like—why is that so clear to you?

Dash: So yeah; let me walk through the logic. I’m sorry. It’s obvious to me, and I’ll tell you why. LLMs on their own, you could implement a million different ways, right? So the tech itself could have been deployed as a tool that I could control as an individual, as a worker. That could be sort of, well, implemented like a spreadsheet is, right? Like, this is this tool that I’m gonna activate on my own to solve a problem in this context.

And the ChatGPTs of the world are sold as subscriptions. They are enterprise tools by design. And they’ve always been designed for being very aggressive about the way they do data retention, and all these other things where there’s an extremely strong bias toward enterprise use. And, very obviously, that’s a business model.

And so what you have is like this dream of either We’re going to make the one worker so much more efficient that we can lay off all of their co-workers, or We’re going to use this as the bludgeon where we say, “You’re going to use ChatGPT to make yourself 10 times more efficient, or we’re going to lay you off.”

And so there’s been this real sort of implicit threat attached to almost all the mass deployments of these LLMs. And there are not, for example, reporting tools or connections into the tools whereby people are able to sort of say, Look how much more time it gave me to think, right? Of, like, variations, right? So if you say the classic scenario—people are like, “I can use this to come up with marketing copy,” right? Like, “I’m good at marketing copy. I’m a good writer. Therefore, I have so much time freed up to think of more concepts, because ChatGPT helped me be more efficient.” Or whatever tool; you know, any of these tools. Like, that could be the advertising campaign for these tools, if they were trying to preserve jobs or be centering workers, instead of management, and be sort of pro-labor.

Warzel: Mm-hmm. Yeah. Right.

Dash: They’re very much not. Right? And so, the thing that I think of—particularly for coders—now, there are times when Claude Code or whatever generates slop code. They certainly didn’t pass. They’re getting better. But for a lot of people, like a weekend coder or whatever, a lot of the experience of coders is: LLMs are freeing you from the drudgery to let you focus on the creative part. Whereas in all the other creative disciplines—like, I’m also a writer—LLMs take away the creative part and only leave the drudgery for you.

Warzel: Right.

Dash: So artists and writers and illustrators, they’re like: “I hate LLMs, because they’re putting us out of work, and they’re only leaving us with the misery.” And the reason that coders are like, “Everybody should love this” is they’re like, “Great; I get to do the joyous part.” And so a huge part of the cultural tension around these things is everybody advocating them is like, Why wouldn’t you love this? And everybody whose industry is being destroyed by them is saying, like, You are immiserating us while you’re putting us out of work.

And I think that part of the disconnect is: Very few people live in both worlds. Like there’s not a lot of people who are, you know, a screenwriter and a coder. Or whatever, you know, whatever two examples you want to point to. And so I think that’s a huge, huge part of the disconnect. And the crux of it is about this labor part. But the thing that’s changing now is: Half a million coders or people in tech roles in the tech industry have been laid off since ChatGPT came out, you know, a little over three years ago. And so now, people are starting to understand there’s common cause between labor in tech and labor in all these other creative industries. And hopefully, people can see they’re all in the same boat.

Warzel: So this is actually a great way to get to, I think, the last part of what I really want to talk about here. Which is the idea that this isn’t the inevitable way that all this has to go. And actually, I really struggle as someone covering this stuff about it. Like, whenever I try to step outside of that box of the top-down “This is the implementation; this is how it’s gonna go,” I immediately get hit with the open source. Yeah, that’s great; that’s awesome. That is maybe how this stuff should work, right?

Dash: Theoretically, yeah.

Warzel: But what are you gonna do? And yet, I just keep being really interested in … let me put it this way. I think that there is a way. Unlike with, let’s just say, social media, right? Like you bought into the Zuckerbergian paradigm of the world, right? And then you sort of realize what we have sacrificed for that very naive version of “Connecting is a universal good.” But there’s something about joining Facebook, you know, which is—it’s like the frog in the boiling pot, right? It seems fine to just join a social network. Like, it doesn’t seem like you’re doing a crazy thing.

With the LLMs, I feel like there actually is this possibility for meaningful and sustained backlash protest. Like, there is a sense of these companies could be the dog that caught the car in a way that I don’t know pertains exactly the same to the social-media revolution, right? Because like, people do … like you were just saying, 500,000 tech workers laid off since ChatGPT. If people do feel these effects, if people do feel the change, if people do feel like, This technology has been foisted on me. You know, everything is a nail when you have the hammer. And, uh-oh, I’m a nail too—

Dash: Yeah.

Warzel: —there could be a meaningful backlash. Not to say it’s going to happen, but there could be. And so, there could be this sense of, for the first time in a long time, the “this is not inevitable” movement could have some purchase. What does that look like to you? What does that movement look like to you?

Dash: There’s a couple of parts. So first of all, the temperature is so much higher, right? The anti-inevitability movement is so much stronger, and the backlash is so much stronger. You know, 10, 15, 20 years ago—when we would push back against social media’s inevitability—people did not give a damn.

Now, if you mention you’re using an LLM, there will be people that are going to shout at you. And, you know, “It’s drinking all the water, and it’s using all the power.” And all this, right?

And they may not be particularly specific or cogent or dead-on in all the criticisms, all the time. Or, you know, maybe intellectually fair all the time. But directionally, they’re correct. Right. Like, these are tools that are harming people and certainly run by people that are not responsible all the time. And so, you know, it makes sense. So I think that the social power behind resisting is so much higher. Especially like, you know, rising authoritarianism supported by the people that run these platforms. There is a pushback. So like, that’s really key.

Warzel: You’re talking about, too—just as an example of this—OpenAI president Greg Brockman made a $25 million donation to the pro-Trump PAC MAGA Inc. So that just being an example of that rising authoritarian.

Dash: Yes; yeah. Right. Thanks. That’s a really clear articulation. And so yeah, but that’s a perfect galvanization of people being like, Okay, I don’t want to pay a subscription to that company at that moment, for that time. And Tressie McMillan Cottom was talking about, you know, people are really feeling it’s important to resist that “inevitability” narrative that these companies are pushing around LLMs.

And the thing I want to do is sort of complicate it. Because I think the challenge—the thing I say about this sort of tech workers’ view of these as normal technology—is that a lot of the people who are resisting feel like, therefore, you say “No LLMs.” And I don’t think that will succeed. Nor do I necessarily think it even should. And that’s informed by our failures in the social-media era. Because when we said, “Facebook is the wrong approach; it’s bad,” and for a lot of reasons, people took that to mean “no social media.” Or when we said Twitter had its shortcomings, people said “no social media.” And that didn’t work.

If I say there are AI platforms that are enabling harms like that toward children … rather than the way to resist the inevitability of those platforms being “Don’t use any LLMs, ever,” say, “Okay, what would it take to have an alternative I feel good about?”

Okay, think about what could a good LLM be. “I want it to be environmentally responsible. I want it to have been trained on data with consent. I want it to be open source and open weight, so that technical experts I trust have evaluated how it runs. I want it to be responsible in its labor practices. Want it to—” Come up with a list, right? So there’s, like, four or five things. And if I can check all those boxes, then I could feel responsible about using it in moderation. And it’s only implemented in apps that I choose to have it in—not forced, like the Google thing where it jumps in front of my cursor every time I start trying to type or whatever. Like, that could be useful. And then I would feel like I was engaging with it on my own terms. That doesn’t feel like science fiction. That feels possible.

Warzel: Just to tie it together with: I really like that vision. That is the vision of all of this that sounds desirable to me. And I look at it up against the new rounds of fundraising from OpenAI, from Anthropic, just from the Meta and Google and xAI of it all. I look at it, you know, up against the idea of these companies IPO-ing, you know, in the next year or so. Raising these huge valuations. And I look at it in probably most importantly, the implementation from the corporate-enterprise managerial level. All of these pressures, all of this movement. The loudness of it.

What you are describing is something that is organic, that is quiet, that is thoughtful. We had the resonant-computing folks on the podcast like a month or two ago.

Dash: Yeah, they’re wonderful.

Warzel: You’re explaining something that is resonant in theory. It just very broadly, like, I mean—do you actually think that that can happen? Like, that we can build this? ’Cause I get so pessimistic about it.

Dash: A hundred percent. Yeah; I get the pessimism. I understand it. And it’s justified. The things I’d say, first of all: Those things don’t have to fail for this to succeed. Like, I don’t think OpenAI goes away. I don’t think you have this David-and-Goliath moment. I think the people who are troubled by these, folks who are the most rabidly against big AI, are like, There ought to be a law, and we’ll have a regulatory intervention. And I’m like, I got bad news for you. That’s not happening in the United States.

And so that’s part of why I want there to be an alternative. Because there’s not going to be what there should be. You know, it’s like these tools are hurting children; therefore we should stop them. Unfortunately, that’s not going to be the case.

But like, how many people on TikTok right now are lit up about the impact this has on marginalized communities, where the power plants are being built, right?

Every single one of them wants this alternative to be built. And so, like, I just like that as a movement. And then you come up with your little seal, you know, your blue checkmark that says, “This is not the world’s worst AI. And if you have to use an LLM, use this one.”

And part of it, for me, is like … having been around a long time, it seemed insurmountable, you know, at one point that people would use a web browser that wasn’t Microsoft’s. Okay. So, yeah. So it’s not easy. It’s not likely, but is it possible? One hundred percent.

Warzel: I think that’s a good, and honestly hopeful, place to leave the conversation. So Anil, thank you so much for coming on Galaxy Brain and talking through the hype, man. There’s a lot of it.

Dash: Despite it all, I remain hopeful. Thanks so much for having me.

[Music]

Warzel: That’s it for us here. Thanks again to my guest, Anil Dash. If you liked what you saw here, new episodes of Galaxy Brain drop every Friday. You can subscribe to The Atlantic’s YouTube channel or on Apple, Spotify, or wherever it is that you get your podcasts.

And if you want to support this work and the work of my fellow journalists at The Atlantic, you can do that by subscribing to the publication at TheAtlantic.com/Listener. That’s TheAtlantic.com/Listener. Thanks so much, and I’ll see you on the internet.

This Is What It Looks Like When Nothing Matters

2026-02-14T07:00:00-05:00

More and more, it seems, I pull to refresh a feed or open up a new browser tab and encounter something that makes me feel as if I’ve sustained a head injury.

Recently, the culprit has often been the federal government. The Department of Homeland Security is putting out white-nationalist dog whistles on X. President Trump posted a video depicting Barack and Michelle Obama as apes. The subtext of every egregious shitpost from the administration is the same: These people are in charge now, and the old rules don’t matter.

A great deal of what I find myself scrolling past exudes a threatening, almost anarchical aura. Just before New Year’s, my timeline offered murmurings of a livestreamer who appeared to have run a person over with his Cybertruck. A week later I would come to know this man as the 20 year-old “looksmaxxer” who goes by the name Clavicular. He hits his face with a hammer to strengthen his jawline and pals around with the white-supremacist streamer Nick Fuentes. Last month, the men were recorded in a club—with other charming manosphere personalities, such as the alleged sex trafficker Andrew Tate—enjoying Ye’s song “Heil Hitler.” (Tate has denied the allegations against him.)

[Read: Why MAGA likes Andrew Tate]

This was trolling, but it also sent a message: “We have enough status and influence to literally get them to play fucking the most—like you can’t even find the song on a single platform,” Clavicular said on a stream after the fact. Brazen displays of status and influence are, of course, what influencers have always been about, but something different is happening with Clav, with Trump and DHS, and with so many other, nonpolitical accounts on social media today. “The reason for the tariffs is the same reason Clavicular hits his face with a hammer,” Aidan Walker, an online-culture researcher, told me recently. “It’s to get attention. It’s to mobilize the base; it’s to prove a point that there’s no rules anymore.”

Social-media platforms—and especially X—have loosened their grip on moderation at the same time that AI tools have allowed for the easy proliferation of slop; never before has there been so much cynical, cruel content and trolling. When Clavicular records himself breaking his body, spouting the N-word, and reveling in anti-Semitism, he’s participating in what Walker dubs “nihilism by default,” an ideology where “the only sources of purpose or profit are the self and the social media machine.”

[Listen: The manosphere breaks containment]

This dynamic is everywhere now. It exists in political memes and propaganda. It drives broad swaths of popular culture. A kind of post-ironic fatalism that was once endemic to seedy message boards has bled into the broader culture, changing how people communicate. Nihilism is now the lingua franca of the internet.

Not so long ago, the most toxic elements of the internet bubbled up on 4chan, a forum that was filled with self-consciously transgressive posts and media including revenge porn, offensive cartoons, and lots and lots of slurs. As Dale Beran wrote in his 2017 history of the site, “4chan defined itself by being insensitive to suffering in that way only people who have never really suffered can—that is to say, young people, mostly young men, protected by a cloak of anonymity.”

Beran further explored the topic in his book, It Came From Something Awful, pinpointing how trolling became a game for disaffected kids with nothing better to do than kill time online. 4chan mutated from a site called Something Awful, which launched just before the turn of the millennium: “90s nihilism endured well into the 2000s, longer than most youth cultures,” Beran wrote. “Like wine turned into vinegar, it could decay no further. Both culture and counter-culture taught new generations to be so wary of being deceived and manipulated, it was best to hold nothing in your heart at all.”

In the mid-2010s, Steve Bannon, then the chairman of Breitbart News, began to see these people as a viable political constituency. The trolls weren’t just telling bad jokes; through Gamergate, they began to understand that they could attract mainstream attention, generate outrage, and mobilize like-minded people. Their preferred version of reality could be foisted onto anyone. “You can activate that army,” Bannon told the journalist Joshua Green in 2017. “They come in through Gamergate or whatever and then get turned onto politics and Trump.”

[Read: “All we wanted to do was play video games”]

This culture didn’t care about old norms and institutions, obviously—in fact, it actively tore them down. Understanding this is essential to understanding the MAGA political project: You can see it in DOGE’s attempts to strip bare the federal government, for instance. But it is also key to understanding other elements of chaos in our culture: This same logic drove the “memestock” moment back in 2021, when Redditors banded together to inflate GameStop’s stock price and manipulate hedge funds into bad positions. (One fund was forced to take a $2.75 billion bailout to survive.)

Today, you can feel the same thing throughout online culture. You can see it in the way that 9/11 memes have become normalized. This week, I saw an altered version of a famous image from that day—an agent whispering into President George W. Bush’s ear about the attacks—captioned with the words “Sir, a second frat leader has brutally frame mogged Clavicular” (a reference, I explain with deep regret, to a viral clip of a jacked Arizona State University student stepping into Clav’s shot on stream).

The process of abstracting huge news events into memes has become much faster. In July 2024, when Trump was shot in the ear, ironic memes popped up just minutes later. (One of the first ones showed a photo of him bloodied, surrounded by his security detail, with the caption “do NOT get your ears pierced at Claire’s.”) Similarly, moments after Charlie Kirk was assassinated at Utah Valley University, an eyewitness recorded and posted a TikTok video that went instantly viral. “It’s your boy, Elder TikTok!” he shouted. “Shots fired!” Before signing off, he asked viewers to subscribe to his page. His knee-jerk reaction—to use the shooting of Kirk for content-creation purposes—was a sign of what would come. Within days, memes featuring Kirk’s face were plastered everywhere in photos—onto celebrities, rap albums, the World Trade Center on 9/11, video-game characters, FBI Director Kash Patel, the faces of random people. The process became known as “Kirkification.”

Arguably, peak Kirkification came in November, when people on social media discovered “We Are Charlie Kirk,” a song—possibly AI-generated—by the artist Spalexma. The over-the-top ballad was paired with countless AI edits and lip syncs on TikTok, Instagram, and X. One of the most popular videos featured an AI-rendered J. D. Vance belting the song to a packed concert hall. By mid-November, the song had hit No. 1 on Spotify’s viral-songs chart, both in the U.S. and globally.

Ironic nihilism appears also to have motivated Kirk’s suspected assassin to etch a bunch of niche memes onto the casings of his bullets. In the days after Kirk’s death, politicians, law enforcement, and many media outlets tried to parse the meaning of the assassin’s inscriptions—to find a motive or assign blame. But there was little meaning to be found. Kirk’s suspected shooter, like many other modern killers, was performing for an imagined audience. Shortly after the shooting, he allegedly texted his partner, “The fuckin messages are mostly a big meme, if I see ‘notices bulge uwu’ on fox new I might have a stroke.”

[Read: Something is very wrong online]

This is what I mean when I use the word nihilism to describe this phenomenon: Memes flatten any event, shrinking it down and making it indistinguishable from the rest of the slop and ephemeral content in a person’s feed, and they also become their own motivation for action. The 4chan logic that turned even the most hideous news and ideas into empty entertainment pervades everything on the internet now—more proof that lol, nothing matters.

This, of course, brings us directly to the Epstein files. On January 30, the Department of Justice released the latest tranche—millions of pages of searchable documents devoid of context or explanation. A constellation of screenshots spread via journalists, influencers, and random accounts. Also, ample fakery: For every genuine Elon Musk email I saw in the files, I saw another that had been fabricated to appear more damning than the originals. On X, Infowars’ Alex Jones posted an AI-generated image purporting to show New York City Mayor Zohran Mamdani and his mother hanging out with Jeffrey Epstein when Mamdani was a child. (The photo was obviously fake; Mamdani’s mother, the filmmaker Mira Nair, was mentioned in one email in the Epstein files, which suggested that she had attended a party at the townhouse of Ghislaine Maxwell.)

Like Kirk, Epstein has become an all-consuming meme across platforms. Long AI-slop videos of Epstein show him as a kind of conspiracy-theory Forrest Gump, influencing every bit of recent world history. In one clip, Epstein is visited in his jail cell by … Charlie Kirk, who gives him a Monster energy drink, which causes him to enter a portal to a snowy fantasy realm. In all of the renderings, Epstein is depicted as more handsome than he was in reality—a typical AI glow-up—and some edgelord types seem to have become enamored with him. “I’m tired of pretending he didn’t have aura,” one alt-right Groyper account posted on X recently.

If the Epstein files were intended as an act of government transparency and accountability, then it seems that they have mostly been a failure. Extensive redactions and scattershot releases have not shed much light on the most heinous of Epstein’s crimes, and the files have not resulted in any arrests or legal consequences here in the United States. Nor have Epstein’s victims gotten the justice they deserve.

[Read: The most ████ administration ever]

What’s happened instead is even more corrosive: The public has gotten just enough of a look into Epstein’s life to see that he remained influential, connected, and even seemingly respected long after becoming a sex offender. The world has gotten a glimpse of the fawning, skeezy shamelessness of his famous hangers-on, but not enough to criminally implicate them. The files have become yet another data point suggesting a deep rot inside of many American institutions. The result again is a pervasive nihilism, where the truth of what’s being discussed matters less than the fact that it is being turned into content, the reaction to which will also become content.

The memes, the trolling, and the Epstein video slop are all a cultural defense mechanism amid a crisis of impunity. The files are more proof that elites of all persuasions seem plenty comfortable saying the quiet part out loud or engaging in egregious, shameless behavior, banking on a culture that has given up on demanding consequences. When faced with evidence of the worst kind of sexual deviancy and conspiracy and no consequences, who could blame a bystander for choosing to hold nothing in their heart at all?

A convincing argument I’ve seen for the prevalence of nihilism, especially politically, is that younger generations have mostly known only political and economic dysfunction.

Five years out from the GameStop ordeal, you can see a similar dynamic shot through the economy, in cryptocurrency speculation and graft, in vapid meme coins and an obsession with gambling and prediction markets. These elements—described as financial nihilism—are especially prevalent in younger generations, who feel that the path of predictable progress (homeownership, access to a thriving job market out of college) no longer exists. “Faced with that reality, taking a gamble on Fartcoin or betting how many times Elon Musk tweets in a week can feel strangely rational,” the Gen Z economic writer Kyla Scanlon wrote in The Wall Street Journal.

Scanlon further argues that this kind of nihilism is “an attempt to find personal agency in a system that’s increasingly denied it to them.” Online, young people wrap their humor in so many layers of irony that figuring out what they’re talking about can seem like a wild goose chase. The posters want people to research and agonize—it only serves to drive more attention to the joke. And so you get nonsensical memes (“six-seven”) and brain rot as Oxford University Press’s word of the year in 2024. The president of the publishing house’s dictionary division justified the choice by noting, “There’s a sense that we are drowning in mediocre experiences as digital lives get clogged.”

[Read: The case for brain rot]

American politics has been a nightmare since before the youngest voters were even conscious of politics. The youngest members of Gen Z were 3 years old when Trump came down the golden escalator to announce his run for president. “Why is Gen Z so obsessed with looks? With wealth? With online virality/attention?” the writer Jasmine Sun posted recently on X. “These values were learned & inherited from the literal most powerful man in the world.” Sun also argued that the broken, sclerotic nature of American political institutions means that the very concept of saving or restoring democracy can feel almost foreign. “Gen Z didn’t experience what there was left to save,” she wrote.

It’s a tidy explanation, but also somewhat believable. Trump’s rise was a signal to many that shamelessness is a market inefficiency in the 21st century—a superpower. That logic was adopted by opportunistic influencers and shock jocks in the initial Trump years. Platforms rewarded it handsomely with attention, money, and power. Many people have grown up and internalized that lesson.

This is not to say that influencers such as Clavicular deserve no blame for their choices—only that they have come of age in a media environment where fealty to an audience and online performance seem so second nature that they adopt the logic of the attention economy by default. Making content becomes one’s only belief structure. The Claviculars of the world are outliers, sideshow performers, which helps explain their (often fleeting) popularity. But this type of nihilism is contagious, and it leads only to destruction: Clavicular breaks his body, while the Trump administration breaks the norms of democracy.

Our culture hasn’t yet been fully subsumed by nihilism, but you can also see it everywhere in different forms: in the mass shooters who seem to care about nothing other than performing for others online. In the influencers Photoshopping themselves into Epstein-file photos to get likes or promote their SoundCloud account. In the overnight viral sensations who become brands and try to hawk a predatory meme coin. In the Super Bowl ads for gambling apps. In a culture of AI slop and brain rot, and in an administration that prioritizes propaganda and graft over governing. It threatens to rip us apart for good if we let it.

Is AI Ruining Music?

2026-02-13T13:00:00-05:00

Updated 4:30 p.m. ET on February 20, 2026

Subscribe here: Apple Podcasts | Spotify | YouTube

On this week’s Galaxy Brain, the host Charlie Warzel dives into the state of the music industry, where streaming economics, algorithmic discovery, and generative AI are reshaping how music is distributed, as well as what it means to make music in this environment. The episode traces how playlists and opaque recommendation systems have left many artists feeling like they’re battling an algorithm. With AI-generated songs now flooding platforms, and even in one case landing on a Billboard chart, the episode examines how automation, impersonation, and synthetic “diet music” are crowding into a system already strained by low payouts and creative burnout.

Charlie is joined by Stu Mackenzie, the front man of the prolific Australian band King Gizzard & the Lizard Wizard, to talk about making music in the algorithmic age. From embracing bootleggers to pulling its catalog from Spotify, Mackenzie explains how the band has tried to protect its creative core while the industry transforms around it. Charlie and Stu explore whether we’re witnessing a normal technological shift or something more existential—an era where music is treated as pure commodity.

The following is a transcript of the episode:

Stu Mackenzie: This ship has, like, well and truly sailed. I mean, it is totally wack to be able to train the algorithm on artists’ work. Totally wack. Like totally cooked.

Charlie Warzel: I’m Charlie Warzel, a staff writer at The Atlantic. And this is Galaxy Brain, a show where today we’re going to talk about music—making it, the future of it, and the ways that technology has complicated that future quite a bit.

Throughout the last decade I’ve been fortunate enough to meet and interview a bunch of musicians across a bunch of genres and levels of fame. And inevitably, the conversation always shifts toward streaming. You’re probably familiar with the basic gripes: Streaming has atomized a musician’s catalog, prioritizing tracks over albums. The economics stink for artists. Musicians have to get big—like really almost Taylor Swift big—to make big money from the streamers. And in order to get big, musicians now need to play the platform game: the same one that creators and average Joes posting anywhere online have to play. Getting put on Spotify or another streamer’s curated playlists is crucial, but so is navigating whatever proprietary algorithms the streaming service might be employing to surface music for listeners.

And so the result has been the kind of frustration you hear from creators everywhere—it is this feeling of having to shadowbox an algorithm to fight for scraps of attention. And it is here where the conversations tend to get, honestly, pretty dark. The musicians I’ve spoken with describe weird things happening—an experimental song from the end of an album, that doesn’t really sound a ton like them, blows up because it got picked up by an algorithm—and that makes it so the streamer lumps their band in with a genre they rarely play in, and then it is harder for their best work to get discovered. Artists who have viral success on one track can find themselves trapped in the same sound, and they’re just chasing the algorithm, hoping to strike gold again. They describe a pressure to churn out new songs and albums faster and faster every year. And there’s this creative confusion that many record labels don’t have answers for. Here’s what one musician told me back in 2024: “Nobody knows what matters, and it’s just like wandering in the desert.”

And the streaming climate has only gotten more precarious since then. With generative AI, it’s now possible to create entire complex songs with a text prompt, or just by humming into your smartphone. Major record labels, like Warner, have recently partnered with generative-AI music companies, like Suno—whose CEO, Mikey Shulman, said in an interview with the 20VC podcast this year, “I think the majority of people don’t enjoy the majority of the time they spend making music.” Now, in the interview he’s referring to the tedious bits of engineering work, but he clearly sees that automating the creative process is a good thing. “It was described to me that we are the Ozempic of the music industry—everybody is on it, and nobody wants to talk about it.”

As it turns out, all that diet music—it’s going somewhere. It’s flooding onto the streamers. For the last few years, across genres, synthetic music has been crowding out human-made music. Chances are, especially if you listen to instrumental music, you’ve unknowingly streamed some smooth jazz conjured by a bot. In late September, an AI-generated song under the [artist] name of Xania Monet became the first to debut on the Billboard radio chart.

There’s other issues, too, like impersonation. The website Rest of World reported last fall, for example, that somebody cloned the voice of reggaeton singer Bad Bunny—perhaps you’ve heard of him—and created a song that reached a top 100 ranking temporarily on Spotify in Chile. That was, before it was removed from the platform.

All of this brings us to King Gizzard & the Lizard Wizard. For the uninitiated, King Gizzard is a popular and prolific Australian band known for bending genres, ripping live shows, and for putting out dozens of albums in a short period. (For example, in 2017 they released five albums in one year.) In July, the group took its music off of Spotify in protest of the Spotify CEO Daniel Ek’s decision to lead a nearly 700 million–dollar investment in a German company that makes military drones and AI-defense tools. “Can we put pressure on these Dr. Evil tech bros to do better?” That’s what the band wrote on their post on Instagram.

But, as the newsletter Platformer reported last year, some King Gizzard fans noticed that several tracks were still available on Spotify. It wasn’t King Gizzard’s actual music, but a kind of ringtone, Muzak-style cover of their songs with similar artwork and information attached. It was only when people clicked on the track that, according to Platformer, it redirected to a different page for a different instrumental-cover artist. The fake King Gizzard songs reportedly had more than 10 million streams. When alerted, Spotify got rid of the fake songs. But the debacle illustrates what artists are up against. Staying on the platform can mean having to play this unwinnable game. But leaving has its challenges, too—losing discovery and, perhaps, having to fend off squatters.

Here’s where I should note that we reached out to Spotify about the King Gizzard situation and AI impersonation in general. A spokesperson acknowledged that “AI is accelerating problems that already exist across the music industry, including impersonation and fraud.” The company shared a series of new policies it announced in September, including “stronger rules requiring artist authorization for vocal imitation and standardized AI-disclosure credits developed with industry partners.”

They continued: “Bad actors can sometimes exploit gaps to push incorrect content onto artist profiles. We are testing new prevention tactics with distributors, investing more resources into content-mismatch review, reducing review times, and allowing artists to report mismatches even before release.”

A spokesman also countered a claim you’re going to hear later in the episode, which is that Spotify pays worse than other streamers. “Every other streaming service pays less than Spotify. Spotify paid out $11 billion to the music industry last year,” they said.

That’s Spotify’s perspective. But what have streamers done to music? What does it mean to make art and music online in 2026, and how can people stay sane and navigate this ecosystem? I figured there was no better person to talk to about this than the front man of King Gizzard: Stu Mackenzie. Stu joined me recently from Melbourne at 3 a.m. my time to help me understand if we’re all, in his own words, “cooked.” Here’s my conversation with Stu.

Warzel: Alright. Stu, welcome to Galaxy Brain. Thank you so much for joining me from Australia.

Mackenzie: What’s up? Yeah, welcome from the other side of the world.

Warzel: It’s summer there, right? You’re, like, actually living the dream there instead of the crushing fatigue of winter that we’ve got going.

Mackenzie: It is a dream, and it is always summer in Australia. Not really, but you know, it is if you believe it is. It is if you’re living in the dream.

Warzel: Love it. It feels like it. I want to start out just by asking: We’re kind of a similar age; you’re an extremely creative and artistic person. What’s your relationship to the internet and technology? Is it an engine of creativity for you? Is it a hindrance? Some of your past interviews about recording—it suggests like even some of that, that you did was a bit analog in nature. But what is your relationship to internet technology and all that as an artist?

Mackenzie: I’m probably going to give you an answer which is probably typical of someone who’s 35—i.e., me and people of our gen—in that it’s incredible. And that it has allowed me to have access to so much music, but also art and culture and everything that was just not accessible to people of 10 or so years before us. And you have to pay that. You have to just go, Fuck, this shit is amazing. You know, like, I hate my phone. I want to fucking throw this shit off a bridge. I want to throw it every single day. At some point, I want to throw my phone into a body of water. But I also just think it’s incredible. You know, like look at this thing—what this shit does is amazing. We’re talking to each other, you know, on the other side of the world, like pretty fucking seamlessly. That is cool.

Warzel: I’ve heard in interviews you say when the band was really starting, it was, like, MySpace days. Obviously things look a lot different now. And I was wondering if you could trace a little bit of that change for me and how it feels from your perspective: evolving as a band and just having the ability to reach people and share your music changing so much. How has that looked and felt to you? Net positive? Sort of net negative? Where do you come out on that?

Mackenzie: I’ve probably got a perspective on this that I’m not sure a lot of my sort of peers or buddies would have the same one as, to be honest. Because for us—the band, I mean—King Gizzard had started in this kind of beautiful, low-pressure scenario. It started in this way of it being a side project, and it worked because it was easy, actually. Not easy in a work sense, but easy in an it fit sense. We did what we wanted to do. From the beginning, it was baked into the whole idea of it, because people didn’t feel pressured to be in this band. It was this weird side-project thing that people weren’t supposed to be paying attention to. So it let us kind of just do whatever we want and not worry about it. But I think when for whatever reason, some people started to support the band, and it kind of got to a point where it felt like maybe we should sort of put some more time and effort into this thing. I think we probably quite sensibly realized that that was actually quite a seed, or DNA, to what the band was. And I think we’ve been pretty good at holding onto that.

And I say that because I think, for the most part, we operate exactly the same as we did even when we had a MySpace. It was kind of just like: We don’t engage super deeply with social media. We don’t create content. But we don’t do anything that’s tough. We just sort of make albums at the pace we want to, and we put out music at the pace we want to. And we still work with the same people, the same sort of small team for the most part. We just have kind of not stopped doing. And I think a lot of the things around us have changed. We’ve tried really hard to stay the same on the inside. And I know that that is so lucky, and I know that we’re very grateful to be able to do that. And it’s weird, but I do truly, like—we’ve grown as people individually in our own ways, and the six of us in the band have grown up with each other and while doing this thing. But when we’re together, it’s still the same. And I’m quite protective of that.

Warzel: That’s really lovely, and something that I think is really rare. But also, I think that’s a good vein for this conversation, in a way: to talk about how you’ve gone about protecting that a little bit, or how you’ve gone about using that to your advantage as a band in this information hellscape, to some degree, or just like chaos. And so I’m curious: Can you talk to me a little bit about the bootlegger stuff, which to me feels really of the spirit of the internet in the good way, and what that is exactly. And why you decided to start doing that—not just like giving the music away, but giving people authorization to press their own albums, sell copies, and things like that.

Mackenzie: Explaining the King Gizzard version of bootlegging is weirdly hard, I think. Let me give it a go. So basically, quite a few years ago, we noticed that people were sort of making, I mean, what I would have described at that time as bootleg copies of our albums. And this was at a time where a lot of our early material had only ever been pressed physically one time: at the very beginning when we released those earlier albums. In quite a few of those first ones, we’d never toured outside of Australia. So we might have pressed 200 or 300 copies, and they had all sold to our friends and our friends’ friends, basically. And then we started to become an internationally touring band and doing all these bizarre things that we never thought were supposed to happen to us. Then, you know, all these other people, and all over the world were like, Hey, can I get a copy of them? One of those records? And we’re like, Oh, really? We’re making this other stuff now. I don’t know. We just didn’t really prioritize looking backward at all. And so I suppose because of that, people started building their records. And then concurrently to that, people started taping our shows. And that was also just a deeply strange and foreign concept to see that, because I didn’t grow up with that culture in any way. I just thought that was super weird. But I suppose I started thinking about that a lot, and how I kind of just actually thought that was very cool. I thought the bootlegging thing was always very cool because we—I never had a problem with people doing that—because they weren’t accessible anyway. I was just happy for people to be listening to the music. And at the beginning it was like, Why don’t we just give them a few albums that they can just do whatever with? Why don’t we just let them do whatever with a few albums? And that’s what we did. We also did this album called Polygondwanaland, which was sort of a prog-rock studio album, I guess.

[Music]

Mackenzie: And we released five albums in that year. And it did start to feel a bit, sort of, capitalist of us, maybe against some of our core values to continually ask people to buy our shit.

Warzel: Five albums in a year is a lot, yeah. I get that.

Mackenzie: Yeah, exactly. So the fourth of the fifth, which was Polygondwanaland, we just made for free. You know, we thought, Well, we should really make this free. Like really free, you know, like this is not just a free download. Like, this doesn’t belong to us. It was like: This is free. You know what free is? This is free. So it was kind of like, you can put it on your movie soundtrack. You can put it on your—you can put it anywhere. Whatever you want. You can take it to the pressing plant and make 1,000 copies and start a record label. You know, you can do anything. This was back in 2017. Looking back after a year of that or something, it was like all these record labels had started up. And all this, sort of, a community of people sprung up around this album. And it did feel like—it just felt so beautiful. It felt kind of like creating life. And it just felt so kind of contrasting to so many of the annoying things about being in a band, where you have to kind of sustain yourself and you have to be a business. So that’s just so annoying. Fucking play music. And it just felt so in line with all of that. But yeah, now we do—just tons of our music is just bootleggable. Just do whatever you want. You can listen to it or download it for free or do normal-people stuff. Or you can do crazy-people stuff and make record labels and shit.

Warzel: And people have, right? Record labels have come out of that, right? It’s wild.

Mackenzie: Yeah, it’s like … it’s so cool. It’s amazing. Yeah.

Warzel: So what I think is interesting about that, is: I want to talk to you in a minute about some of the Spotify stuff. Some of the stuff that has been hard and squeezing artists to some degree. And some of that all comes from this idea, right, of: You’ve got to find and build community in whatever ways that you can, right? Get the music out to people, but also have something there. And I feel like what you guys have, whether you stumbled upon it or whether it’s just, you know, the way that your ethos as a band and as human beings have decided to be creative and make art, is that it seems like you’ve been able to bypass a little of that commercialization and form that community. Which then, in turn, is sustaining, right? Like, I think about Phish, the [Grateful] Dead, you know, the idea of that sort of taping community, right? And this idea of like, We’re going to play lots of live shows. People are going to go, and we encourage that taping. That thing gives, as you just said, gives us the ability to deliver a different kind of show every night—that creative challenge of improvisation and locking in and all that stuff. But at the same time, too, it creates this ecosystem, right? And the ecosystem can then become as obsessive as you want to be over shows and archives and “last time played” and “first time played,” and all these different types of things. And it builds the lore. And it fosters that community in a way that I think ends up being probably way more generative, right, than just like, Yes, stream our album. Here it is. And then here’s the next one. Right?

Mackenzie: Yeah. That’s my worst nightmare: to kind of be, like, just constantly trying to push the thing. That’s not fun. No shade to anyone who wants to operate like that, but it doesn’t work for me. It’s a personal preference, I think. And I do feel extremely, extremely fortunate to be able to take the kind of risks and the wild swings. We take tons of falls, too. But we’ve been able to weather it. I’m very grateful for that. It does also—and I know you want to talk about Spotify stuff and everything—but it does sort of make me feel like we have a bit of a duty to do stuff like that. Because King Gizzard exists in this weird realm of “It exists because we kind of took risks the whole time.” It hasn’t been, sort of, despite that. It’s been because of that. And I think we are just luckily and beautifully placed to just be ourselves. And I kind of feel like we have a duty to do that, in some ways, because I know that in so many ways it is easier for us to do it than for other people. And so now, with something like Spotify: Like, it doesn’t pay the bills anyway. For the most part, we’re kind of a touring band, and we make records that people buy IRL. Like, I don’t know. We sort of just exist in our own sort of—we’d whirled over to the side.

Warzel: Right. So let’s just talk about the Spotify stuff, since we’re there. So last summer, you all made a decision to remove the music. It sounds like Spotify CEO Daniel Ek leading the investing of this military-drone company was, I guess, the last straw. You mentioned earlier just that this wasn’t something that’s necessarily paying the bills. Can you walk me through the decision? In terms of what else, in your experience with the platform, led to this, you know?

Mackenzie: It’s a pretty evil corporation, for the most part. I mean, and to be fair, so are a few of the other streaming services as well. And it’s actually kind of easy to hate Spotify, to be honest, as an artist. Like, they pay way worse than almost all the others to begin with, and they’re the biggest. So they’re kind of setting the standards in so many ways. I feel like all my mates were already saying “Fuck Spotify” constantly anyway, before we did it. And before all of the military-investing stuff came up, as well.

It does sort of feel like when you’re making music, making music is—maybe this goes without saying, or maybe I’m just going to say it—it’s such a vulnerable thing to do. It’s actually, like, really hard to make music with other people. I mean, again, maybe that does go without saying. It’s hard sort of emotionally, and it’s hard sort of in a way of—it’s hard to motivate yourself. And it’s just this vulnerable sort of like, weird—I don’t know—sort of thing that we and other musicians constantly put ourselves out there in this weird world to do. Because we believe in it. I guess a lot of the things happening around Spotify started to feel like—it honestly just made me and the others like not going to work, right?

And I love work. Like, work to me isn’t this bad word. Work is awesome. Work is great. I want to go to work, you know? I want to make stuff. And all this shit made me not want to go to work. And, you know, we had a lot of conversations with a lot of other musicians who I love and admire, and some of my other mates left Spotify. I just thought, Man, that’s … I don’t know. For some reason, it wasn’t something that I really thought of being possible. It also sounds like you just spend so much time thinking about music and working on the music, and Spotify is the platform that most people listen to it on. And then a few of our mates left. And I just thought—I didn’t even really think about that as an option.

Warzel: I’ve been having this conversation for a couple of years with a number of artists and musicians, especially. And I was talking, to prepare for this, to another musician. Who’s honestly done really well for themselves, very popular, and mentioned that the Spotify stuff—especially the algorithmic-discovery stuff—had just ground them down so much creatively. Like, basically had said: I—use the same terms of work, right—I consider this to be work. I consider when I got to get through, you know, like I’m problem-solving, right? Like, it’s not all fun and games. It’s like my job sometimes. And I’m not afraid to grind through it and do that.

And then there was this feeling recently that they weren’t able to stop. Which was like, I’m going through all this, like, preparing this beautiful meal, right? And I have this total worry that this thing that I have no control over is just going to stop people from coming to my restaurant. Right? Even if they like the food all the time, they’re just going to forget that the restaurant exists. Right? For some kind of weird reason. And this idea also that songwriting is for them this craft: this thing that they love to do, right? And that it’s become sort of … it’s gone from the skill game to this slot-machine game, right? And if you’re thinking about the casino: Really good blackjack players don’t want to spend time at the slots, because the slots are stupid and they’re arbitrary, right? And that’s how it felt fighting for people’s attention. It was a really interesting feeling, though. That it’s got that similar thing of like—it’s just making coming to work feel juiceless, or bad.

Mackenzie: Yeah. And that’s interesting to hear you say that, because I do think there are a lot of analogous sort of—that’s happening in so many industries at the moment, in different ways. And I feel like I can kind of talk about it from my perspective of being a musician. I feel like I have conflicting ideas about it. I feel like I almost open my mouth to say something and I stop myself, because it’s hard to sort of form a fully thought-out opinion on a lot of this stuff. Because it’s still happening. And, again with technology, there are certain elements of a lot of the AI or generative things that I look at, and I’m like, Fuck, that is so cool. That is fucking amazing that that shit is possible.

Warzel: To democratize some of this. To give people some of the tools, some of the ability to say, like, I see this, I feel this, and I want to make this for myself. Right? That is like going back to the first principles of all of this stuff, the way that humans did it before it was professionalized. And yet, if you do reduce it to: Hey man, I want to do this, and I can do it now because I can write out a prompt and push a button, that’s fine on a creative-explosion scale, right? It’s not fine probably in the sense of: I wanna use your guys’ style of music, and I wanna fit it into a Spotify playlist and then, you know, just ride off the back of you guys no longer having your music on Spotify. Right? Because, like, that’s an example of something that’s happened to you guys, right? You have people who are kind of like, yeah. People are coming in, and—

Mackenzie: That really happened. And I totally agree. And it’s something that I’ve talked about, as well. And I think where I landed was: We are doomed. It does feel like we are fucking doomed when that shit happens. It’s like, what do you do about that? This ship has, like, well and truly sailed. I mean, it is totally wack to be able to train the algorithm on artists’ work. Totally wack. Like totally cooked. Totally fucking horrible.

And it’s just—I can’t believe that that has just happened. And it’s like, Whoops, didn’t mean to do that. Shit, okay. Oh well, that happened now. Let’s just move on. What? That’s crazy. That’s why it feels like we’re doomed.

Warzel: Can you describe for me how it happened? How you found out about this? Because essentially, right, what happened was: You guys took the music off of Spotify, and people were just coming in with, like, ringtone-style stuff. Is that how it went down?

Mackenzie: I don’t know how people are making this shit. It’s like, honestly, when I listen to these King Gizzard, sort of, AI-artist songs, they’re insanely funny to me. Because it’s so dark and so twisted and so strange. And it’s so weird that it’s happening to us. All I can do is laugh. It’s insanely funny, but in the most twisted and dark way. It’s just … my reaction is to laugh, but it’s very dark and very weird and very 2026. But basically, to anyone that doesn’t know, we took our music off Spotify. And then, tons of music has just been—it’s probably there now—stuff has been taken down. More songs come up. And I suppose, because our artist page must still exist, or maybe you can still search for it, I don’t know—but it means that if someone is looking for King Gizzard, it’s very easy for them to find this stuff. And actually, there is a very weird problem where you can actually upload music impersonating another artist and go onto their official profile very easily. That’s a whole other thing. It’s super easy to impersonate another artist and end up on their official profile. And this has happened to tons of artists.

Warzel: Have you talked to Spotify about this at all? Or have people on your guys’ behalf?

Mackenzie: Yeah, quite a few times. And they’re—quite a few times—and they are always just like, Oh, whoops; we’ll take it down. And they do take it down. But, you know, there’s more up there before you know it. And it’s not just us. It’s happening to tons of artists. It’s a story, actually. I think it’s just exacerbated this whole thing, because it’s left a hole. So when I’m talking about these, like, AI King Gizzard tracks, they are on our profile. Like, it’s King Gizzard. It’s not someone else. It’s, like, “us,” that’s the weird thing—but it’s not us. It’s just … I don’t fucking know. It’s just—yeah, it’s weird.

Warzel: But what’s so interesting is you all leaving Spotify. It sounds a little bit like, you know, it hasn’t been that hard for you guys. Is that correct?

Mackenzie: I suppose so. And I would want to preface it with, you know, I don’t see us as necessarily being a model for other people to follow. Because I know that we exist on a very unusual path, and I’m very proud of that. But I’m not sitting here saying … I don’t want to sit here and say, “Everybody should leave Spotify.” And if people want to do that, that’s cool. But I’m not here to pontificate on anything, actually. That’s not my vibe.

But for us, we have had nothing but quite beautiful press around that, and around leaving Spotify. And I think quite a lot of people have discovered our music. Because with all of this, the hardest part about leaving Spotify is making your music inaccessible to so many people who listen to music only on Spotify. And that doesn’t feel good. Like, I don’t want to do that. Nothing about that is what I like about what we’re doing. But it felt like the right price. And that’s okay. And I am proud of doing it. But it does feel like people [are] maybe taking notice of what we’re doing, as well. Yeah; it’s cool.

Warzel: I wanted to just end with—it’s easy to get bummed out about some of this stuff when you’re talking about technology and art and all that. Is there stuff that makes you feel optimistic? About—maybe you can just speak for yourself and not for the world and art and creativity. But for you going forward, do you feel like some of this technology, some of this stuff, some of just the explosion of information that you can access, that that’s going to be generative for you, as you continue to evolve as an artist?

Mackenzie: I remember when—like a few years ago now, and I’m sure there are many people listening to this who have had a similar experience—when ChatGPT first came in. And a few people were like, You should try this thing; it’s like pretty freaky. And I was like, Okay, cool. And I started writing in just the most batshit prompts, like, You are a grain of dust traveling through the fucking cosmos. Explain to me what you see along the way. Stuff like that. That just felt so new. And there was a time when I was like, Wow, this feels so inspiring. I feel like this is actually doing something new. I’m sure this is what it felt like when someone first used the thesaurus to write, to help write poetry or something. Imagine if you’d never used a thesaurus, and then you just picked up one. And you’re like, Whoa; this is going to make my writing so much easier!

Warzel: The game has changed!

Mackenzie: Yeah, right? I was thinking about a lot of things like that. All these things we kind of take for granted that were new technology at some point. But I don’t feel very interested in that now.

Warzel: There’s an interesting interview that Brian Eno did earlier, I guess last year, talking about this kind of stuff. Brian Eno is someone who’s created so much generative music and so much ambient out there, and pushed the boundaries of this electronic music and stuff. He was talking about playing around with ChatGPT and this idea that when you first start using it, there is that feeling of, Whoa, like: This did something different. And then he was saying the more times he started to do it, the less interesting it got to him. And he had this great word for it, which is, like—he used to do watercolor painting. And when he would put the brush in the water between the different things, no matter what happened, no matter what colors he was using, the water at the end of the thing would be this weird sort of maroonish-purple. And he called it—munge was the color. And he was like: That’s ChatGPT. The output is munge. Like, No matter what I do, it always kind of comes back to this kind of drab sort of soulless mix of all the stuff. And I thought that that was kind of like a perfect summation of maybe why that kind of stuff is not all that interesting to you right now.

Mackenzie: Maybe. Yeah, that’s a very funny take. And yeah, I definitely relate to that. I’m not sure I ever got really, really, really deep, or I really got to know ChatGPT’s personality. But I do know that it has a personality. And I do know that it’s what, like, a good percentage of the world is using. And that—to me, as an artist—is not interesting to engage with right now. At least not in the kind of music that we’re making.

That moment that I was talking about—when I first did use ChatGPT, and it kind of did blow my mind—it actually felt niche at that time. It felt like something that my parents, for instance, wouldn’t have known about. It was at that point, whatever that point was—a lot of it does come back to kind of what I was saying earlier. And I just want to do what makes me want to go to work. And what makes me kind of excited to just get out of bed and race into my studio and make music. Like: That’s what makes me want to pick up the phone and call one of the other guys in the band and say, Be at the studio at nine o’clock in the morning. Like, Let’s go; I’ve got these ideas. I just want to put myself in that head space.

I wonder if this is like when the drum machine came out? Or, I wonder whether it is not like that at all. You know, when the drum machine came out and people thought, I miss real drummers. I miss real drummers, with all that beautiful feel and all their imperfections. And, you know, I could tell the difference between this drummer and that drummer just by hearing them on record. And all those things are true. And then, you know, so much amazing music came out of that. And a ton of this wall of senses is drum machines.

Warzel: I think it’s all a cycle, right? That’s the only way that I can kind of keep it all in my head, because I think we all as humans tend to do that “It’s so over; we’re so back; it’s so over; we’re so back” kind of thing. And instead of being completely doom and gloom about it, or dismissing it as nothing, right, I think it’s a part of this cycle. I think there—as with something like a drum machine—you get to do different things and manipulate it in all these different capacities. And kind of push the bounds of something, right? And there’s a lot of creativity and interesting, like, wall-breaking there. And at the same time, too, I think you sacrifice. You pay the price of a little bit of that humanity, or a little bit of that, you know, spice-of-life type thing. And then people, if you go too far in one direction, eventually I think people start yearning for the other thing.

What I find really inspiring about what you guys have charted out for yourselves is that it feels like there’s a real focus on the humanity part of it. On the human element, on like the drive for the new thing. The creative push. And I think that that is what a lot of people are craving more and more these days.

Mackenzie: Yeah; that’s interesting. And I appreciate that. I think with what we have done in our career, in a lot of ways, we have not been very tactical about anything. And we have made decisions from the gut; we definitely have just tried to prioritize doing things in real life with real people. And that’s, yeah—the obvious stuff, like playing shows and stuff like that. But also just, like, meeting people and talking to people and just being a real person. Like, I would pride myself on being pretty ordinary in a lot of ways. And I think that’s cool.

Warzel: As an ordinary person, I totally agree. Man, I so appreciate this conversation. And, you know, grounding it in the humanity and the real-personness of all that. I think we need more of that. So I appreciate all of this and the insights. And all the time, man.

Mackenzie: Cheers, Charlie. I am grateful for the chat, and the reflection and the inwardness. And the, yeah—it’s just good to talk about this stuff before we go insane.

Warzel: That’s the whole point of this podcast. Talk about it before we go insane. Not so we don’t go insane, but just before we do it, you know?

Mackenzie: So we understand why. Yeah.

Warzel: Thanks again, man. I appreciate it.

Mackenzie: Cheers, Charlie. Thanks, mate.

[Music]

Warzel: That’s it for us here. Thanks again to my guest, Stu Mackenzie of King Gizzard & the Lizard Wizard. If you liked what you saw here, new episodes of Galaxy Brain drop every Friday. You can subscribe to The Atlantic’s YouTube channel or to Apple, Spotify, or wherever it is that you get your podcasts.

And if you want to support the work that I am doing, and the work that all of my colleagues at The Atlantic are doing, you can subscribe to the publication at TheAtlantic.com/Listener.

Thanks so much. See you on the internet.

This episode of Galaxy Brain was produced by Renee Klahr and edited by Dave Shaw. It was engineered by Dave Grein. Our theme is by Rob Smierciak. Claudine Ebeid is the executive producer of Atlantic audio, and Andrea Valdez is our managing editor.

This transcript gives proper attribution to the 20VC podcast for Suno CEO Mikey Shulman's quote. An earlier transcript attributed the quote to The Guardian; the audio and video versions of this podcast still do.

The Manosphere Breaks Containment

2026-02-06T14:30:00-05:00

Subscribe here: Apple Podcasts | Spotify | YouTube

On this week’s Galaxy Brain, Charlie Warzel takes listeners deep into the internet’s fever swamps to examine how figures who once would’ve stayed on the fringes now dominate mainstream feeds. The episode charts the rise of Clavicular, a young livestreamer who’s gone from an absurdist curiosity to a fixture in the manosphere and its adjacent right-wing influencer culture. Using Clavicular as a lens—his extreme body modification, relentless self-documentation, and a willingness to do anything for attention, Charlie discusses the rise of nihilistic Zoomer influencers. Then Charlie is joined by the internet-culture researcher Aidan Walker, who helps situate Clavicular alongside figures such as Nick Fuentes and Andrew Tate, revealing how the “looks-maxxing” movement collides with nihilism, grievance politics, and an anti-political, “algorithm-first” ideology. Together they explore what happens when the gatekeepers are gone, and when nihilism becomes a default way for budding attention hijackers to build an audience.

The following is a transcript of the episode:

Aidan Walker: Their very existence is proof of something not working.

And so, in a way, their project is to exist, to be seen, to be popular. That’s why he’s going to say the N-word on stream. That’s why he’s going to read the humiliating text from his father on stream. It’s a total commitment to that project. Because I think his existence just sort of proves that the gatekeepers are gone.

[Music]

Charlie Warzel: I’m Charlie Warzel, staff writer at The Atlantic, and this is Galaxy Brain, a show where today we are going to expose ourselves to a lot of really awful internet content so that you don’t have to. We are really going to plumb the depths of the online fever swamps here. Or maybe it’s more appropriate to say that the characters from the depths of the online fever swamps moved up in a rather concerning way from the fringes all the way into popular culture.

Over the past few months, a very young livestreamer named Clavicular has risen out of obscurity and become part of a stable of manosphere influencers and right-wing online figures. Now, if you’re a normal person who is not chronically online, you probably don’t have the faintest idea who Clavicular is. And I’m sorry to ruin that for you, but bear with me.

Pretty much everything about Clavicular is preposterous. One of his first brushes of internet fame was when this photo of him went viral on Reddit. He was flexing in the mirror, and it was taken by this old, disinterested lady. In December, he made headlines for livestreaming on Christmas Eve while apparently running a man over with his Cybertruck. In interviews, he’s claimed he’s hit his face repeatedly with a hammer to crack his jaw and sculpt it.

He said he’s done methamphetamines to get hollow cheeks. He seemingly takes all kinds of steroids and meticulously documents all of it.

Clavicular is a “looks-maxxer,” and that’s part of this online subculture that is obsessed with going to extreme lengths to achieve a chiseled-faced notion of perfection. It’s an online community that’s gained a fair amount of notoriety in recent years, in part because there’s this overlap there with these other online groups that all cater to disaffected and vulnerable men. It’s always a little bit difficult to categorize these groups, but I think it’s safe to say that in Clavicular’s case, he’s somebody who represents this slice of the manosphere—this big group of popular influencers who traffic in blatant misogyny, online nihilism, and all kinds of destructive trolling behaviors.

Now, historically, somebody like Clavicular would be strange enough and arguably off-putting enough that he might toil on the fringes for quite a while and slowly build up this audience and influence in some of these backwater like-minded communities. But instead the opposite has happened. Clavicular has blown up extremely quickly.

He’s been palling around and collaborating with these manosphere influencers like Andrew Tate, Nick Fuentes, who is arguably one of the most significant media figures on the far right. As we’ll discuss, Clavicular went viral in part because he seems to be willing to do or say absolutely anything and associate with absolutely anyone in order to be famous.

This is how he ended up in a club in Miami a few weeks ago with Andrew Tate and Nick Fuentes singing along to Ye’s song “Heil Hitler.”

Clavicular’s rise is pretty interesting in part because he represents this edgelord, manosphere influencer who also has a pretty incoherent politics. He hangs out with Nick Fuentes, yes, but he’s also called J. D. Vance fat and ugly and subhuman.

Now, I know what you’re thinking: There’s arguably a little bit of risk in taking all of this deadly seriously. In part because Clavicular and all these guys are really young, and also there may not be a lot going on up there. Clavicular himself said on a stream recently: “All I think about is content. I’m sorry, bro. Like, dude, I literally only think about content. You’ve got to understand.” End quote. Good stuff.

And yet, I think Clavicular’s rise—and the rise of these influencers that he’s associating with—may actually reveal a good bit about the direction of online culture in a world where Nick Fuentes and these nihilistic Zoomers seem to be gaining influence online. It’s really easy to dismiss these guys when you watch them online. It all just seems so bankrupt and vain, and often racist and sexist and just kind of devoid.

It’s a worse version of the 2016 pro-Trump shock-jock ecosystem that was led by now mostly forgotten influencers like Milo Yiannopoulos. But this is not the media ecosystem of 2016. And it’s not just Fuentes, who has reach, ending up on Tucker Carlson’s show and vexing Republicans who’d rather not have to side with him or disavow him.

Take Nick Shirley, the 23-year-old YouTuber whose video on alleged fraud at Somali American day cares in Minneapolis went viral late December. A lot of Shirley’s claims had been debunked, and Shirley is by no means a serious journalist. But people in the Trump administration listen to that. The media circus online around his video was used by the administration to justify the surge of ICE agents in Minneapolis.

In other words: Shirley’s provocation, which is content that is made just for the algorithm, broke off the internet into the real world, leaving violence and chaos in its wake. Things right now that can seem trivial or beyond the pale or just so stupid in the second Trump administration often aren’t.

So who actually are these influencers? Where do they come from? And crucially, how seriously should we be taking them? Joining me today is Aidan Walker; he’s a writer and internet-culture researcher.

He’s currently the new-media manager at the Carnegie Endowment for International Peace. And he spent years documenting the evolution of memes and online culture at the site Know Your Meme. When it comes to content creators, he knows of what he speaks, because he runs a popular TikTok account himself. And Aidan is just the perfect person to situate Clavicular, Fuentes, and his cast of characters into a broader context of online and political culture. He joins me now.

Aidan, welcome to Galaxy Brain.

Aidan Walker: Happy to be here.

Warzel: So we’re gonna walk through this. This is gonna be some hand-holding, I think, in general, because you are a student of the fever swamps, of the meme craziness, that I think a lot of people do not ingest to the same degree. And I’ve been in those swamps myself. I dip in and out of them.

But I wanted to do something here where we really kind of walk people through all of this. So I want to start—let’s just hope that people haven’t had the pleasure—but who is Clavicular? I don’t even know if I’m saying his name right. Who is he?

Walker: Well, okay, to start from the very beginning: Clavicular is a dude named Braden. He started posting on the internet as a teenager. So when he was around 15, 14, he was on these looks-maxxing forums, which—some of it overlaps with 4chan. Not all of it is 4chan. But he’s out here talking about body modification, essentially.

Warzel: That’s what looks-maxxing is for people.

Walker: Yes.

Warzel: Gotcha.

Walker: So maximizing your looks through various methods. And for a while he’s posting, you know, like many teenagers on social media, gathering a bit of a following. And in the past two months—essentially, I mean, I first heard of him in December, and as you said, I spent a lot of time in the fever swamps—he kind of vaulted to this level of viral prominence. Mostly for stuff on Twitch that he was doing, and Kik actually. So livestreaming his actions, his adventures, his misadventures.

Warzel: So tell me a little bit about what you know; he started doing this when he was much younger. What was he doing in that community? Like, tell me what the looks-maxxing community, what do they—do they believe anything? What do they believe? Tell me about his involvement and that community.

Walker: So the thing to know about looks-maxxers—and I want to be careful with my words, because the lore here is deeply complicated and controversial—but there’s an adjacency to incels. The core of it is kind of this philosophy that the only thing in life that actually matters is how attractive you are. Everything else is sort of a scam. Everything else is a lie.

Women only care about how good you look, men only care about how good you look, and grades don’t matter. Nothing matters. And so the rational person, faced with this situation, will do anything they can to look better.

And so the looks-maxxers will do everything from steroids to, kind of famously, hitting themselves in the face with a hammer or another hard object. The science is that it’ll make microfractures. “Science” in scare quotes, of course. And that’ll reform your facial bones so that you look more like a Chad. You have a sharper jawline. And they’ll also do things like “mewing,” named after a Dr. Mew, which is the sense that you can kind of like—

Warzel: It’s like structuring your jaw, right? Like biting down on something, or doing basically jaw exercises to work it out—as you would if you were curling your biceps or something like that.

Walker: Yeah, yeah—to make it sharper. To make it more muscular. And in a way, it kind of ties into this pseudoscientific health stuff we see with the RFK [Robert F. Kennedy] Jr. BS that goes around. But it’s very much like these young, alienated men on the internet doing anything to make themselves look better. And then kind of posting constantly about the insane lengths you’ll go to, like hitting yourself in the face with a hammer to look good.

Warzel: So what did Clavicular do in this? Did he have a pretty normal trajectory on this? Did he break off from this? He just sort of reached people’s radars late last year. But was he someone of prominence here in this world, as far as you know? Or was he someone who just kind of ascended late in that?

Walker: Yeah, it’s funny that you use the word ascended, because that’s the word they use to describe going from having a soft jawline, looking like a virgin, and then you ascend to look like a Chad. So Clavicular—prior to kind of popping off in November, December—had sort of transferred over to posting on TikTok, and doing kind of the classic posting that attractive people do on TikTok. You know, like dancing, kind of like Blue Steel–ing the camera, like Zoolander. But then including all these “ascension” content—where he’d have a picture of himself when he was young and pimply and then one now, where he’s kind of sculpted and hot.

And so he was prominent in the niche, but he was far from the No. 1 kind of person there. And as to what he did to become prominent: He hit a dude with his Cybertruck during a livestream, and he told a conservative podcaster that he supported Gavin Newsom over J. D. Vance because Gavin Newsom is a 6’3” Chad, and Vance is fat and ugly.

Warzel: So it’s a lot to take in. But what’s interesting about that is simply the degrees to which one can become famous now. Also, the degrees to which somebody like that can kind of pop out of nowhere for the most ridiculous things. To your mind, how has this idea of looks-maxxing—again, it’s not something a lot of people were talking about even, you know, two or three years ago. How has that ascended? What is its rise been as a culture?

Walker: So looks-maxxing is very old, like probably 10, 15 years old. And some of kind of the earliest very active meme subcultures on the internet are like bodybuilders and then incels. And not that every bodybuilder online becomes an incel. It’s a funnel; it’s a pipeline. Not everybody goes down it, which is important to remember. The looks-maxxers aren’t all, like, fallen lost boys who can’t be saved.

Warzel: Who are other looks-maxxers who’ve kind of broken through in this? Or is Clavicular the first that you feel like has achieved this kind of Twitter main-character level of fame?

Walker: So I think Clavicular is the first that has achieved this level of main-character fame. And I think one reason for that is: There’s definitely been people who maybe I’m not aware of that have gotten big on Instagram for doing this. There’s been looks-maxxing influencers for years. But the roots of it, really, are like in this sort of anonymous image-board kind of culture where people aren’t necessarily hustling for the fame. And I think Clavicular’s innovation is that he kind of married that looks-maxxing niche—which is very strong, very vital, has been for years—with sort of this general niche of, like, “I’m a young, hot person” on TikTok. I go to the club; I go out and make videos of myself dancing; I talk about hitting on girls. And so he’s kind of married the two together. And to my mind, I think, he’s one of probably the first really big, famous looks-maxxers to kind of break through into a category of more general fame. Which probably says something about the derangement of our society.

Warzel: Yeah, I think probably. But you have Clavicular, and we’re kind of working toward something here where Clavicular is this character to come out of this corner of the manosphere that is very extreme. About doing extreme, sometimes, damage to your physical body in order to look good—but also is part of this incel, kind of disaffected male culture that has a political valence. Would you say that it has a real political valence, or would you say that it doesn’t really?

Walker: So it has a political valence—but it doesn’t map very neatly onto right-left. So I think part of the reason why it was shocking that Clavicular endorsed Gavin Newsom over J. D. Vance is that anyone on the internet would expect a dude like that to be kind of right wing, in part because it’s kind of incel adjacent. But more than anything, there’s this kind of nihilism to it. Like, the reason he’s endorsing Gavin Newsom isn’t because he is against the [Trump] administration or wants a certain policy outcome. It’s because that cold logic of looks-maxxing says the only thing that matters is the sharpness of a guy’s jawline.

Nothing else in the world actually applies. There’s no morality. There’s no rationality. It’s just, you know, What degree is his canthal tilt?—which is the way your eyes are in relation to the bridge of your forehead. And so it’s a total evacuation of all the other things that people care about. And just a total replacement of it by, you know, (A) How do you look? Is it good? Does it fit the metrics? And (B) How does it perform on the platform? And so if it has a political valence, I don’t think it’s like Clavicular wants a strong social safety net. I think it’s like: Clavicular believes the way that everybody else talks is stupid, and that he’s going to performatively just insist upon this on-the-surface bizarre, and yet strangely cohesive, system of just describing the entire world—describing all human relations—by reference to how people’s faces look. So it’s a total lack of belief in anything social. Anything beyond just, like, “Oh, that guy’s hot or not hot” as a metric for evaluating something. So it’s anti-political, maybe, in a way.

Warzel: So, okay, so we’re gonna put a pin in that. Then there’s another character in this universe. They’re all going to join together, in the worst possible Avengers way. Tell me who—and I think plenty of people will be familiar with his name—but tell me who Nick Fuentes is.

Walker: Yeah; so it’s interesting. Nick Fuentes and Clavicular, after Clavicular’s glow up, have started to collab and still are. Fuentes is about my age, born in 1998. He is a Gen Z, white-nationalist political commentator who first got big around 2019. He has a group—they’re called the Groypers, after a version of the Pepe meme that’s sort of an unappealing toad. And his main platform now is: He does these livestreams where he takes questions; he kind of impersonates a news anchor, and he’s always drilling down on these, like, Steve Bannon–esque points. But even a bit more like explicit and further out. And he kind of got to the scene in 2019, by starting this Groyper war against Charlie Kirk and Turning Point USA—where Kirk would go around campuses and do these events, have people asking questions, do these debate bro–type things. And Fuentes and his Groypers would come into the audience and then ask these questions that would push Kirk to take two steps further to the right than he already was: to endorse something blatantly anti-Semitic or blatantly racist, when Kirk might’ve just been doing the dog whistle before. And so that’s kind of how he soared into prominence as like the furthest stretch of the alt-right. And now, post–Charlie Kirk’s death, Fuentes sort of got mentioned a bunch more. And his stream has kind of been continuing all these years, and now he’s kind of the big bad boy of the online right.

Warzel: Yeah, he went on with Tucker Carlson. He’s sort of, he’s taken his shots with other people. And it’s just been sort of, on the right, like—there was a big blowup with the Heritage Foundation and whether or not to, you know, basically attack Tucker Carlson for having Nick Fuentes on. And this idea that the Heritage Foundation wasn’t going to weigh in. And it was this feeling that Fuentes is always pushing just what he was doing, in the quote unquote “Groyper war” with Kirk, right? He’s always trying to push, even with his presence. People have to disavow or not disavow him. And he does have this meaningful constituency of Groypers. What do Groypers believe? Do they believe anything?

Walker: Good question. I think within the Groypers, there’s probably a lot of diversity of belief. But I think they can kind of be understood as a subset of the alt-right, kind of the more radical edge of it. I think that most of the—not to generalize—but most of the animosity that they feel is probably channeled toward minorities, toward Jewish people, but also toward the center-right: like the Charlie Kirks of the world, kind of the establishment. Most of their viral clout has been gotten by Fuentes being the dude that these people say, “That’s too much; this guy’s too crazy.” And then all of the most crazy, disaffected young people will say, “Oh, really? The guy who’s too edgy for Charlie Kirk? That’s right for me.” And so I think they’re just kind of into pushing that edge.

And Fuentes himself has a bit of like a white-nationalist type. You know, solidarity politics in a way. It’s hard to even call it solidarity of, you know, saying to young white men or to others of like, This is your group; go with this; oppose everybody else. And so, to the extent that they do believe something, it’s kind of just like being racist. And I think one thing you really start to notice when you spend time in the fever swamp, and you study things like green-text stories on 4chan, or just these various memes, is: I think a lot of these men who are into Fuentes are marginalized, for lack of a better word. They’re disabled; they’re neurodivergent; they come from a poor background; they come from a familial-abuse background. And a lot of the content is about that. It’s about that kind of—

Warzel: So that’s what they’re confessing in those threads, you mean? Like, you’re not surmising this? This is actually what they’re saying in their threads.

Walker: Yeah. I mean, in as many words, yeah. I mean, they’re posting threads about, you know, Can’t get a job. I’m a drain on my family. My mom hates me, you know. Can’t get with girls, you know. I’m ugly. I’m fat. I can’t figure anything out. Like, it’s guys who are stuck in that situation. And not everybody who’s in that situation or a situation like it makes that choice, so they should be held responsible. But I think one thing to understand about it is: They choose the Groyper Toad as their emblem. You know, that’s not a jackbooted, strong character. That’s a chubby, off-putting, gross toad. And so, I think a lot of what that movement does—the service it provides—is it allows them to feel, you know, accepted in this freak brotherhood. Where all you have to say to have this sort of all-encompassing, you know, ever-loving group is: You just say the slur. And then you’re on the other side of that Rubicon. It’s like a gang. It’s that kind of initiation ritual. And it seems to solve these people’s problems sometimes. But I think it always leads down very destructive paths.

Warzel: This is interesting too, then, in the convergence, right? Because you have someone like Fuentes. You have a group of these people; obviously not all of them are like, you know, feel like they are horribly marginalized. Some of them are just, you know, edgelords or people who are just, you know, out-and-out, proud racists and love to engage with other racists. But there is this level of disaffected feeling.

And then you have the looks-maxxing community, right? Which is also—there is potentially just as much of a home for disaffected people there, or people who feel, you know, I’m ugly or I’m a drain or I’m whatever and want to optimize themselves in this way, or find that to be really appealing. So this brings us to this video that came out a couple of weeks ago that went pretty viral on a lot of my different feeds—mostly with a lot of people feeling this was like the end times. But it was a meeting of Clavicular [with] the manosphere influencer who I believe has been indicted in different countries for sex trafficking, Andrew Tate, and Nick Fuentes, among others. There’s a few other characters there who we don’t need to get into. They’re at a club in Miami. They are goofing around, seemingly having a good time, and I guess goad somebody in the club to put on the Kanye—or Ye, as he’s known now—song “Heil Hitler.” Which is its own provocation. And the idea is that they’re all dancing to it, but that the whole club is dancing to it. And it’s this sort of moment of total trolling—but also this like out-and-out-proud racism that everyone around, in that clip, seems to be totally okay with in that moment. And I think watching that, so many people felt like, What the hell is happening? And so, I would ask: What the hell happened there?

Walker: Yeah; it’s horrifying. And they’re blasting that song in the rented limo going to the club beforehand.

Warzel: Okay, I didn’t see that.

Walker: So that was, I think, always the plan. And Kanye, Ye, I think is like the representative artist for some of these guys. Because, you know, the verses of that song before the horrible chorus are about how he can’t see his kids anymore, and he’s a drug addict and a washout and a sleazebag and all this stuff. And I think that they definitely staged this moment in the club for maximum shock value—so that it goes across people’s feeds. And I think there’s sort of two sides to it. The first is that, you know, if you transgress in that way, and you get people outraged and shocked, that’s the only play these guys have ever done.

You know, from Gamergate to today, from 2016 to today, they’ve just always pushed the boundary. And then been rewarded with clicks and attention, even if there is pushback. Then I think the other side of it is just this sort of violent show, of like, “Oh, we can get away with it.” Right? Like, this is the one thing that liberals would say you can’t get away with. And just by doing that—having the video go everywhere on X—they’ve said, “Hey, we got away with it.” And they’ve done this summit of all these guys. And Clavicular’s status there is interesting, because he’s the newest one, and he’s the youngest one. And he’s the ringleader of it in a way.

Warzel: Before, and correct me if I’m wrong, but before they went to the club, I believe Nick Fuentes, this guy Sneako, and Clavicular sit down for this livestream, right? You said that this was, like, a very rich text, right, for someone who wanted to understand the differences between these people. You have Clavicular—who is younger, very young—and then you have Fuentes and Sneako, who are in their late 20s. That wouldn’t seem like a huge difference, but in these online-influencer worlds, it actually is. There’s so many cycles in between them of how to think about content, how to think about their politics.

You wrote that it seemed like, in the stream—which I just have to say to people who might not seek this out on their own, it’s so lame. Like, it is just so deeply lame. These guys are like pulling up computer chairs, wearing suits, nice clothes. It’s like, people walking around in the background; they’re in a living room, and they’re just very not charismatic. Sometimes Fuentes comes off on his own show as very charismatic. Here, he’s just kind of hunched in a chair, very uncharismatic, having these conversations that seem sort of brain dead, honestly. Like, they’re racist and whatever, but also just kind of not animated, or very even interested at all. But you write that Fuentes is interested in this stream, or it seems, about describing his political project around this white identity—that there’s this coherence, almost, to what he wants to do. How he wants to sort of take conservatism from the mantle of the Boomers and move into something else. And you write that Clavicular seemed totally uninterested in that. And tell me a little bit about that, and the differences that you saw between someone, you know, sort of older Gen Z and younger Gen Z in that sense.

Walker: Yeah, that was—I mean, it’s a horrible stream to watch. I kind of just inflicted it on myself, I guess, because I have a professional—anthropological, sort of—interest in these spaces. And I think also, just on a personal level, being the age that I am—same age as Fuentes and Sneako—I think you just kind of grow up, and you see a lot of kids in your high school or whatever kind of get pulled down that path. In the same way you see kids get into drugs or something. And it ends up badly for them. And it’s always been this force, just like in the internet. So I think that’s part of the fascination. And where Fuentes is describing his political project, most concretely for me, it’s when he’s talking about getting this gold sponsorship that used to be, I think, Mark Levin’s, who’s kind of a Boomer conservative commentator. And Fuentes is very hung up on getting this sponsorship, because I think, for him, it would mean that he’s beaten the Boomers, and now he’s the mainstream of the conservative movement.

Warzel: This is like a “buy gold” thing, right? Sorry to interrupt.

Walker: Yeah, like a “buy gold” thing, like the classic, you know, alongside like survival-prep stuff that supports right-wing radio. There’s no reason for it to matter to Clavicular, but it still matters to Fuentes and to some extent to Sneako. And so, I think there’s a difference between kind of elder and younger Gen Z, in that for Clavicular—I mean, I’m sure he’s not a nice guy. I think he’s probably a racist. I think that someone who isn’t hateful wouldn’t do the things that he’s done. But he’s not like—

Warzel: He does seem to use the N-word quite a bit.

Walker: Yeah; exactly. But he’s not interested in remaking the state in the same way that Fuentes seems to be. He just seems to be interested in making himself as handsome as possible. In this totally, I guess you could say, like, Randian, self-interested whatever. Whereas Fuentes has some desire to be seen by the institutions, or to be seen beating the institutions. For Clavicular, it’s irrelevant from the jump.

Warzel: Yeah, you described it as “the exchange corresponding to the overall idea of an algorithmic ideology, replacing an institutional ideology.” That is really interesting to me. The idea that someone like Fuentes, whom we would think of as extremely online and using an online movement to some effect—obviously, attention hijacking, growing an audience, etc., still having institutional desires. Right? Still wanting some kind of political power. Whereas Clavicular seems to only want … like, he rejects institutions almost fully. Right? In his sort of incoherent, you know, left-right politics or whatever. But also in the idea that he really only cares about the algorithm.

Walker: Yeah; he only cares about that cold logic of, like, algorithm. How many views can you get for saying this or doing this? You know, how many aura points can you get for mutilating your face in this way? There’s just this total lack of belief in anything that is social, anything that is, you know, institutional. Which, to me, it translates as kind of an anti-politics. But what I sort of meant by ideology—and this is maybe something I do with my content, a little tongue in cheek, and then maybe it starts to become sincere—is, you look at these guys, and you look at this boring, horrible stream as a rich text. And I think what you can sometimes extract from it is: These are people who have this set of distorted values about the world but that have somehow found visibility and attention. And why, exactly, is that? Their very existence is proof of something not working.

And so, in a way, their project is to exist, to be seen, to be popular. That’s why he’s going to say the N-word on stream. That’s why he’s going to read the humiliating text from his father on stream. It’s a total commitment to that project. Because I think his existence just sort of proves that the gatekeepers are gone.

It kind of proves that these weird, sleazebag, disaffected, angry young men—that there’s nothing holding them back. The fact that Clavicular is there is proof that you’ve won. And I think Fuentes, being a little bit older, had this formative experience of the gatekeeper still being there. He got famous by being the guy that was too radical for Charlie Kirk. And so, he needs there to be a Charlie Kirk there. Or some figure like that. And I think Clavicular doesn’t have that need. Which I think testifies to kind of the moment we’re in, and that they’re able to self-sustain just by being this beast of the algorithm.

Warzel: You write—and I thought this observation was really clarifying—that this is nihilism by default. That Clavicular is nihilism by default. And the elder Zoomer, maybe, the Fuentes ideology, is a little bit of a nihilism by disillusionment. And I think that that is really interesting, because one way that I look at the internet, throughout generations of it, is like: The culture’s always layering stuff on top and absorbing the thing that came before it, right? Everything’s a little more of an abstraction, right? Like, the Groyper is an abstraction of the Pepe meme, which was its own abstraction of a thing from, you know, a cartoonist that got popular on 4chan. These things layer on top of each other.

And it’s interesting to think of the nihilism being layered on top, to the point where you get these people who are just like—they don’t know why they have this “lol, nothing matters” disaffected feeling. It’s just sort of like what they saw on the internet from this community and they’re just adopting it. That’s kind of—that’s terrifying, man.

Walker: Yeah. And, I mean, it’s the difference between resenting the idea of adult supervision, I guess, because you think you’re oppositional, or better than the adult. And then the feeling of there just having never been an adult in the room. You know, like I think someone like Clavicular, since the time he was conscious, you know, Donald Trump has been the central figure in American politics.

I think that was even something Clinton said in 2016—that it’s going to be a generation of people raised watching someone who breaks every rule, gets away with everything, and gets wildly rewarded for it. And there’s tons of other figures in American and global life who are not Trump who exhibit that kind of behavior as well. But I think it’s just years and years of that there’s not even any juice to be had in opposing the establishment, because the establishment is so absent from your life, honestly.

Warzel: Yeah; I heard—this is a slight tangent, but somebody who is part of Gen Z was saying to me: The reason why the “democracy dies in darkness” stuff doesn’t resonate with us is because Trump’s been around since the very beginning of this. He’s the only thing I know about American politics, from my own experience. What are you talking about? What is a functional feeling there, right?

And I think that that is a really interesting piece of, you know—if we’re describing the water that a lot of people are swimming in, especially like a much, much younger generation. That dysfunction, that nihilism, doesn’t not feel the same way. But also that the appeal to institutions, or the appeal to, you know, norms or civic virtues or values or whatever—liberalism even—that it’s not on the radar, almost. And that is concerning, I guess.

Walker: Yeah. It’s not even opposed, necessarily. It’s just not legible. And I think it’s important to point out, too, that the majority of Gen Z people are not nihilistic looks-maxxers who want to burn down civilization. It’s just: Some people kind of make that choice, or get drawn to it. But I think there is just overall a sense of the old coordinates—of whatever the old order does—have just been gone for so long.

Warzel: Well, and this is where I’ll—I’m glad you brought that up. This is where I want to try to push back a little on all of this, right? And get a sense of, I would say first off: Can you describe the size of all of this? Because, like you say, it’s not everyone. It is its own., Like, there’s a strange thing that’s happened, where the fringe has become not the fringe anymore. And yet, it still obviously is part of a fringe of some kind of thing. It’s a little hard to understand. But, like: How popular are these guys? How do you think about it? How do you quantify it?

Walker: So, I mean, you could look at subscriber counts and stuff. And my impression is—forgive me if I don’t, viewers can look this up at home—I think Fuentes has around 100K on his Telegram channel, which is a bit more of a closed, intimate social media. Actually more used overseas than in the U.S., I think. But there’s sort of two ways to measure it. I think it’s like you can go for raw overall count. In which case, of course, Mr. Beast is bigger; mainstream influencers are bigger. They have more people. But then there’s the sense of like, as an influencer, you probably would rather have 100 paying subscribers vs. like 100,000 people who will just like your video.

And so, I think some of what allows these fringe figures to throw a disproportionate weight around on the online ecosystem is that once they have a fan, they have a zealous follower. They have a devotee. You know, it’s so radical, it’s so extreme, to get to this position where you’re gonna post and say “I’m a fan of Nick Fuentes.” Like, that guy’s taking up a lot of your time and a lot of your life—in a way that’s someone who’s a fan of a more mainstream creator, they probably aren’t. And we saw the same phenomenon with 4chan: pretty small platform in terms of user base, but has this disproportionate influence on internet culture because of how committed and fiendish they all are. And so, I think these guys are definitely someone that most young people will have heard of. They’re definitely on their radar. Which is where they want to be. But I don’t think they’re at all the dominant influence on people. They’re an option on the menu, which is troubling, but they’re not the main course, maybe.

Warzel: My overarching question here though, I think, is: How much should we be paying attention to these guys? And I know that’s sort of impossible to ask. But there is a sense—I’ve gone so back and forth on this in my own coverage, in my own thoughts with all these different people—where you’re like: You know what? This guy is an out-and-out horrible racist-type person. There’s just no reason to give this any oxygen. And then you see the audience continue to grow, the specter continue to grow.

Then in the right-wing-media ecosystem, they’re platformed by just a big-enough person, say, Tucker Carlson and Fuentes. And then it’s like, The New York Times is writing about him every day, and it’s a person who has to be dealt with now. And then it doesn’t matter whether you think there’s enough oxygen, or you should give them oxygen or not.

How do you think about paying attention to these guys as a culture and a society? Because it’s lame. Like it feels like in a just society, it’d be like, These guys are absolute freaks. We don’t need … like, this is not worth anyone’s time. And yet, that’s not happening.

Walker: Yeah. I think that’s always been kind of, as you point out, the concern with kind of internet-culture journalism and covering these people. Because the thing they love most in the world is, you know, The Atlantic talking about them. Like, that’s the goal in a way, or one of the goals. But I think they’re worth paying attention to, because I think we still often make this distinction between right-wing-media ecosystem and the administration. But it’s the same thing. The DHS account; the kinds of memes they’re posting. The fact that after Kirk died, J. D. Vance went and took that podcasting chair. These guys—in terms of clout, in terms of the way they speak to the base, in terms of the way they inform policy—Ted Cruz sits down with Tucker Carlson.

Tucker Carlson is the more powerful figure there in terms of the right wing. So I think it’s crucial to pay attention to them if we kind of want to understand the way that this administration is consolidating power and thinking about its place in the world. And I think from another angle, one thing you see a lot when you look at online niches is that you sort of have these big whales, and then there’s other creators. There’s like a big delta, I guess, between the No. 1 and then like the No. 6. And I think with the right-wing ecosystem, the big whale is Trump. And then there’s sort of different flavors of that, and all of them collaborate and sort of help each other and then repost and work off of the big guy.

And so I think understanding that, and paying attention to it, sort of tells us more about this moment and that administration than a lot of kind of the analyses that are like, you know, Look what they’re doing to the courts, or Look at this latest order. Let’s think about what’s the reason for the tariffs on XYZ country. You know, the reason for the tariffs is the same reason Clavicular hits his face with a hammer. It’s to get attention. It’s to mobilize the base, it’s to prove a point that there’s no rules anymore. And so, I think that the structure of these radical communities tell us something about what the government itself is doing.

Because the last point that I always sort of try to make—thinking about it generationally—is like John Ganz, the writer, makes this point of Groyperfication within the government. A lot of these young conservative staffers—guys who are like 35 now—10, 15 years ago, they were the kids on 4chan. And that’s the person who’s been empowered. That’s the milieu that’s coming out of. And I think that’s part of why I really pay attention to this. It’s that I find that, you know, it helps to sort of know what’s going on and what’s driving it at a deeper level.

Warzel: Where do you think this goes? Because, again, when we think about the layer cake of nihilism, so to speak—you’ve written about some of these guys that they burn so bright with their trolling and one-upsmanship and content-first mentality that it’s not sustaining, right? They will eventually probably immolate in some way, on a creator level. But at the same time, the culture nudges forward, right? It becomes a little more extreme. Even if those stars in their galaxy kind of burn out.

Walker: I think it goes to death. Both for them personally, and then probably for social media, the internet, the country. There’s no image of the future. It’s just this race to the bottom. Break every taboo, break every rule, destroy your body.

Warzel: You mean death, like, literally?

Walker: Like literally, yeah. If you sort of agree with the analysts who kind of see this moment as fascist in one way or another, I think that’s where fascism leads you. It’s a form of death worship. And I think it’s also, to the extent that what the looks-maxxers do—like, part of why that interested me is that it feels so poetically kind of apt to describe the whole thing as that You destroy your body; you destroy your life chances for the sake of an online public that really doesn’t care about you. Because the incentives are always for them to double down, always for them to do more. The only place it ends is with their self-destruction.

And the question is whether the rest of us go along with that. And I think that in many ways—insofar as the first thing they destroyed was the institutions, and now they’re trying to destroy the rest of us and themselves—it’s like, Where do we stop it? Where do we detach from that project? And I think it’s something that … I don’t know what I’m saying. I’m an internet-culture analyst. I don’t know about national politics, in the end. But it’s like: We just need to find another option. And I think it’s important to tell young people that this project ends in death. This is a guy who’s destroying his life for a livestream. And they’re destroying the country for clicks.

Warzel: I think that’s a great place to leave it. A depressing place to leave it. But also, I would just say—as someone who has covered this stuff for getting close to two decades, and always being like, “This internet-culture stuff, it’s not politics,” and then every year being like, “It’s kind of politics.” So I wouldn’t sell yourself supremely short. We’re not geopolitical analysts, but this stuff has bled into the highest levels of, at least, American politics right now. So I think the perspective is worthy. Aidan Walker, thank you for coming on Galaxy Brain.

Walker: Charlie, thank you so much. It’s a pleasure to be here and talk about a terrible topic.

Warzel: Yeah. Into the abyss together, you know?

Walker: Yeah; better than going alone.

Warzel: Thank you again to my guest, Aidan Walker. If you liked what you saw here, new episodes of Galaxy Brain drop every Friday. You can subscribe to The Atlantic’s YouTube channel or on Apple, Spotify, or wherever it is that you get your podcasts about Clavicular. And if you want to support this work and the work of all my fellow journalists at The Atlantic, you can subscribe to the publication at TheAtlantic.com/Listener. That’s TheAtlantic.com/Listener. Thanks so much, and I’ll see you on the internet.

How to Survive the Information War

2026-01-30T13:45:00-05:00

Subscribe here: Apple Podcasts | Spotify | YouTube

On this week’s Galaxy Brain, Charlie Warzel opens with what it means to live in 2026, when our phones can drop us into graphic, real-time violence without warning—and when documenting that violence can be both traumatizing and politically consequential. Using recent footage out of Minneapolis as a lens, he explores the uneasy collision of algorithmic feeds, misinformation, and the moral weight of witnessing. Charlie also traces how viral documentation can puncture official narratives, pushing stories beyond political circles and even into “apolitical” corners of the internet. Then, Charlie is joined by Amanda Litman, a political digital strategist and the co-founder of Run for Something. They discuss how to be a good citizen in the information war without losing your mind. Specifically: In an age of algorithmic fragmentation and billionaire-owned platforms, does sharing that devastating image or news article actually accomplish anything? Or is it just performative activism? Together they explore how nonpolitical creators and everyday people can be especially persuasive messengers, and how to pair online engagement with offline activism. It’s an episode about how to stay engaged without surrendering your nervous system and how to use the internet as a tool for connection, clarity, and action, not just despair.

The following is a transcript of the episode:

Amanda Litman: I think there’s a hesitation that some people have, especially after the last five years, of like, I don’t just want to post and have, like, performative activism. I don’t just want to do virtue signaling. I think virtue signaling is good. I think it is good to want to show people you’re a good person.

[Music]

Charlie Warzel: I’m Charlie Warzel, and this is Galaxy Brain. For the last few weeks, I’ve been obsessing over a question, the answer to which may be impossible to know: What do we do with this window we have to the world?

We’re living in this moment of genuine chaos in a decade that’s already been full of global conflicts, a botched insurrection, a pandemic, political dysfunction. All of that colliding on an internet full of conspiracy theorizing, synthetic AI slop, endless memes. Our phones bring us delights and numbing distractions, but also fresh horrors every day.

Maybe you, like me, woke up last week and found yourself faced with a series of auto-playing videos of agents mobbing Alex Pretti in Minneapolis. Maybe you watched as an agent shot him while he lay on his knees on the street. Or maybe a few weeks ago, your phone showed you the amateur video taken from multiple angles of Renee Good’s last moments before she was shot by an ICE agent at close range. Or maybe you logged on this fall and had to witness Charlie Kirk’s gruesome assassination in your feed.

Social media has enabled us to vicariously witness terrifying scenes. This is especially true in recent weeks in the Twin Cities. Sometimes we seek them out, but other times they find us in moments when we’re especially unprepared to see them. A colleague of mine recounted checking his phone and accidentally stumbling upon an auto-play video of Pretti Saturday morning while at the aquarium with his daughter.

A reality of being alive in 2026 is that you may accidentally open your phone and witness a man being shot to death and then have to figure out how to continue on with your day. The cognitive dissonance here is extremely difficult to sit with. To say nothing of the psychological trauma of witnessing such violence, even if you’re able to stay away from the worst of it. Watching armed, masked agents pulling elderly people out of their homes in the cold or detaining young children—that takes a toll, especially when you close your device and you find yourself back in your own life. The world feels on fire, and your life is just moving on.

It can feel alienating, and it can make a person feel so angry, sad, and ultimately powerless. And yet, the power of this kind of documentation—the kind that’s being performed by observers in Minnesota every day, who are risking their lives to film clashes between agents and protestors—that cannot be understated.

Pretti’s last seconds were captured from multiple angles in sickening footage that was widely distributed on social media and news organizations. It is able, though, to be seen and dissected online precisely because of the observers who were there to document it, who did not drop their phones when the gunshots rang out, and who kept recording when federal agents piled atop Pretti. All of it has had an immediate effect, in countering the administration’s smears that Pretti was a “would-be assassin” who “tried to murder federal law enforcement.” The news and the footage of Pretti’s death managed to break through the usual informational chaos.

On Reddit and Instagram and Facebook pages, the videos of Pretti’s last moments appear to have galvanized people who don’t normally engage or post about politics at all. You have sports-meme pages and golf influencers. The climbing subreddit, knitting accounts, not-safe-for-work pages. I mean, even a Facebook page as apolitical as the “Gravestones of New England” was calling out ICE’s excessive force in the aftermath.

Something has changed.

All of this underscores that if the truth is ever going to win out over propaganda, it can only do so in the face of overwhelming evidence—the collection of which has become ever more treacherous in the second year of Trump’s second presidency.

You can understand the power of documentation and of witnessing, but you can also struggle with how to consume it all yourself. I’m reminded of this great and sobering essay by The New York Times writer Amanda Hess. It’s about witnessing sensitive content videos on Instagram from the conflict in Gaza: this wrenching series of videos and photos of dead children, all on an app dominated by ads and influencers and algorithmically chosen photos of friends doing fun stuff.

Of that, she wrote, “Sometimes, when I tap on a post from a journalist in Gaza, Instagram suggests next steps. ‘Are you sure you want to see this video?’ it asks. It tries to point me instead to ‘resources’ for coping with ‘sensitive topics.’ It suggests a crisis hotline for disaster survivors and responders, but I am not a survivor or a responder. I’m a witness, or a voyeur. The distress I am feeling is shame.”

What Hess is getting at here in part is how what we might witness on our devices isn’t just traumatizing or radicalizing, but it’s also linked explicitly to the specific platforms that serve this content up to us. And these platforms are incredibly fraught curators. Many of them are owned by billionaires with their own political agendas. For example, late last week, TikTok finalized the sale of its U.S. operations to an American investor group. And that includes, among others, Oracle, the private-equity firm Silver Lake, and the investment firm MGX.

Now, this became relevant over the weekend when Pretti was killed, because influencers attempted to upload videos to the platform, criticizing ICE. And some found that they couldn’t do that. Others got the videos up, but noticed they’d received zero views. And that was very suspect. So many of them naturally took to social media to express their outrage. They said that they’d been censored by the new leadership. “With Larry Ellison’s takeover, TikTok is already silencing voices on the left and anti-ICE, anti-Trump content.” That was what the popular podcast I’ve Had It posted on X on Monday morning.

This is just a perfect encapsulation of this moment, right? TikTok noted that it had suffered a data-center outage and that user uploads and views had been affected. The transfer of ownership from ByteDance, TikTok’s parent company, to the U.S. ownership group, paired with a data-center outage, might mess with the platform. That’s an entirely plausible scenario. There’s also the fact that nonpolitical content on the platform, like posts from the NFL’s primary account, also showed zero views on Monday, which suggests a broader issue here. But others just refuse to believe that. The only thing in this that is clear right now is that there are so many people who are extremely and understandably wary of TikTok’s ownership and seem to assume at all times that somebody’s thumbs are on the scales.

All of this makes the job of being an informed citizen just tortured. Do you stare at the misery machine? If so, how much? What do you take from it? Is the misery machine real life? Is it not real life? “Don’t be a doomer!” they say. “But also don’t underreact!” The truth out there is probably as bad, maybe even worse, than you think. But if you remove yourself from it all, it’s easy to slip back into something that is like complacency. It is crucial to pay attention right now. It’s also crucial not to lose resolve or get sucked in or traumatize yourself. There’s a real world out there. Touch grass, people say. The real world is likely full of people around you who tether you to your community. People you love, and who love you. But the trauma in our screens—that is also the real world.

So what are we supposed to do with this window to the world? It is such an unanswerable and essential question right now. Sometimes it seems like this window has only made people grow apart, more callous, more self-interested, more addled. And yet that very same window often shows us that people are good and decent and that they care about each other and organize and have these huge reservoirs of empathy.

We’ve seen that in Minnesota this month. So with all that, how should you or I—how should we all—think about posting and consuming and being a good citizen in the information war? To answer that, I spoke with Amanda Litman. She’s the co-founder and president of Run for Something, which is an organization that recruits and supports young and diverse progressive candidates running for down-ballot office. She’s the author of multiple books, including When We’re in Charge: The Next Generation’s Guide to Leadership. And most recently, she’s been writing on her Substack about the ways that people and politicians can calibrate themselves to the internet to feel some agency and make a difference in the world.

She joins me now to talk about it all.

[Music]

Warzel: Amanda, welcome to Galaxy Brain.

Litman: Thank you for having me.

Warzel: So you’ve been in politics awhile as an organizer, a digital strategist. And I get the sense from following you online that you’re a pretty online person, and that you’ve been closely following what’s been happening in Minnesota online. And so I want to start by getting a sense from you of what you’ve witnessed these past three weeks—what the experience has been like watching this—but then also what you’ve been seeing in terms of the broad effect of other people watching this collectively.

Litman: It’s been so interesting, because my algorithm very clearly knows who I am. It knows I like politics, it knows I work in politics, it knows I’m a mom. It’s nice to feel seen in some ways—

Warzel: —by Big Tech—

Litman: —recognition is recognition. But the people I follow are, generally speaking, people who work in politics, for the most part. Or people I’ve met through the political circle. Those folks have been tuned in since the beginning. What I have noticed over the last three weeks, both from the folks I follow from other parts of my life, but also from the cultural creators, especially from the parenting creators, from the book people, from the romance writers, from the other parts of my life that are not politics—and I think this has been true across the internet—is people feeling like the shooting of Alex Pretti is the last straw. That this is the thing where you cannot stay silent. And it has felt so reminiscent of 2020 in that way, in that there is a tipping point where the conversation cannot continue as normal. In a way that I have found shocking, to be honest.

Warzel: Yeah; it’s really wild to see it. I’ve seen it too, and I think a lot of people have remarked on this. In my other life, I like to play golf. And so I follow a lot of people who are golf influencers or whatever, or my algorithm gives me people like that. And that is usually a pretty, like, We’re staying out of politics crew of people, you know? And there was one image I saw posted online of a golf influencer. He’s just at a driving rage, just hitting a shot. And he’s like, This normally wouldn’t be a political account, but golf is political because you can’t play it if you get shot by an agent of the state in the street. And it was like this moment for me of, This has bled through in a really, really shocking way.

Litman: It feels like the conversation has shifted from it being a thing that political people are talking about to a thing that everyone is talking about—in a way that doesn’t feel like it’s happened, honestly, since Jimmy Kimmel was fired. Which feels like the last big cultural moment that sort of broke through, at least in the online space.

Warzel: And so you wrote this piece—this is one of the reasons I wanted to talk to you today—about the politics of posting, and this idea of what we do on these platforms, whether it matters. But I think that this is a great context for it—this idea that there has been this shift. And you wrote it, I think, before Alex Pretti was shot, a day or two before. And you started the piece by describing seeing this now-famous photo of this 5-year-old boy who was detained by ICE. In the image, he’s wearing this little backpack and this hat. And it’s an extremely, I think, difficult thing for anyone to see—but especially a lot of parents. It really broke through online. And you talked about posting that image yourself on your own platforms in order to put some pressure on ICE or raise awareness here. And then you wrote, “But: Did my posting (or any of our posting, on any platform) do anything to further that effort?” Walk me through your process there. Did it?

Litman: Yes and no. I’m not sure if I actually wrote this, but I have come full circle here. Used to be, back in my digital-strategist days—so when I worked for the [Barack] Obama campaign in 2012, that was my first job out of college. I was doing online fundraising. I then did digital strategy for the [Florida] governor’s race and then did online fundraising for Hillary [Clinton]. I was kind of, maybe counterintuitively, Posting doesn’t matter. It is important in that it is participatory, but it doesn’t change things. It’s like, It’s not enough. You can’t just post and be like, “Great job; done!”

Over the last now 10 years of the work I’ve done in seeing how things have shifted, I have moved my barometer for success a little bit. I think in this moment, as you have written so much about, there is no singular media source. Everyone is getting a slightly different algorithm, a slightly different, fractured media diet. They’re seeing something a little bit different. So if I post the picture of Liam Ramos or the headlines from the Minneapolis paper, it might be the only thing that someone sees about that story if they follow me. Even though it feels like my entire newsfeed is 100 percent that, it might be the only thing. And there’s actually some really interesting research about this, that I think Wired wrote about in December: how political content posted by nonpolitical creators has more influence than political creators doing it. Like, it does more to move people to the left, in part because of the parasocial relationship that people have with folks who are talking about golf or parenting or books or whatever it might be.

And like, I’m not a creator; you’re not a creator. We’re not influencers in that same way. Although maybe we are, which is sort of a separate conversation.

Warzel: We’re all creators now.

Litman: But that idea that you—as a trusted source for your friends, your family—they know you. They know where you’re coming from. Seeing you speak up about it—“you” in the broad sense—that can be really, really powerful. And it takes the place of having a conversation in person, that most people are a little bit too afraid to have these days.

Warzel: It’s such a great point about the influencer thing. When you look at—I think this is something, and we can get into this a little later down the line—but when you think about how different ideologies have harnessed the media. Like, there was this whole long conversation after the 2024 election of Who is the Joe Rogan of the left?, right? And the smartest people on this subject who study this type of stuff, who really understand these dynamics, all said, You can’t create that. Because Joe Rogan is a guy who is a standup comedian who started a podcast about standup comedy where he interviewed comedians, then got into mixed martial arts and did that, and built this huge audience that loved and trusted him for whatever reason. And then he started to get interested in conspiracy theories, and then politics. And all of the work that you do for a decade of building that trust in that audience, when you do turn on the politics jet—like as you said—this is such a powerful source for people who have that relationship with them. And I think that it’s really interesting to think about with our social feeds. This happening in a very, one-to-one, very kind of community-level aspect. I think that’s really, really smart.

Litman: Well, and I think there’s a hesitation that some people have, especially after the last five years, of like, I don’t just want to post and have, like, performative activism. I don’t just want to do virtue signaling. I think virtue signaling is good. I think it is good to want to show people you’re a good person. And I write this, like, it feels a little self-righteous. It can be a little bit annoying. But one, the right has tried to make vice signaling very cool. They have tried to make being an asshole and a bigot and a piece of shit on the internet the hot, cool thing, and shame you if you’re a good person. I think that’s dumb. We should be proud of being, of doing good things. And it should be okay to post things that signal your values. And I will tell you, as somebody who runs a nonprofit, if you’re making the donation so that you get the receipt to share: I don’t care. That money spends all the same. It doesn’t matter why you’re doing it, as long as you are doing it and actually taking action in a meaningful way that moves the needle forward.

Warzel: You write that it could also be a way to find your people. I think a lot of people don’t believe that. But can you give the case for finding community through some of these platforms and networks?

Litman: I mean, it’s hard for me personally to articulate this, because everyone knows I work in politics. This is my life. But I hear this from the people who engage with this material all the time, which is like—it’s so validating to see someone say something or post something you agree with in a way that might surprise you. Like, I didn’t know that writer felt that way. I didn’t know that creator felt that way. And especially for people in real life, like, I didn’t know that fellow parent at school that I only follow because we see each other at the playground every so often feels similarly. Now we can have a conversation. Now maybe next time I want to volunteer in a campaign, I could reach out to them.

It often can feel either like screaming into the void or preaching to the choir. But at least when it comes to preaching to the choir, you might reach someone who just didn’t feel like they could sing loud enough, and you might encourage them to speak up. And it feels so cringey to say out loud. It really does. And I get that. But in this moment—especially with the propaganda war that the government is trying to instill—it really, really matters that there’s a different kind of dispersion of information.

Warzel: I think there’s something to the cringey thing. I’ve heard other people say this, so this is not a deeply original thought, but: I think, broadly speaking, the whole idea of cringe is really, I think, toxic. The idea of labeling everything as this. In the same way that virtue signaling is good when the world is full of this vice signaling, when there is a terminal irony that is basically poisoning all these different platforms—and the White House posts like they’re a 4chan moderator or something like that—I feel like cringe is good. Bring us some cringe, right? Because it basically just means you care a lot. And I think that that is something that I would like to see more people not be afraid to show. That they’re caring in that way.

Litman: Rachel Karten had a thing, I think it was last week, about like the Mister Rogersification of marketing this year, of how more and more brands and marketing are moving into Mister Rogers vibes in their social media and leaning into this idea of kindness and hope. And I have been thinking about this a lot, even in 2025. Like, my hottest take is: I think that the Democratic nominee in 2028 is going to be whoever can appropriately balance fight like hell with kindness and joy. I don’t know if you ever see that meme that’s like “The people yearn for Glee.” They just want Glee to be on. We want that Glee joy, like the television show. There is something real to that. It is nice to see nice things. Fear and rage, it just burns you out. Whereas when you remember that people care, and they believe that better things are possible, it makes you feel like you can get out of bed in the morning.

Warzel: Not to put you on the spot, but are there good examples of brands or whatever that have done this? Successfully pivoting to the cringe slightly, or the Mister Rogersification?

Litman: Okay, so I’m gonna quote from Rachel Karten’s newsletter, because I thought it was … like it really spoke to me. She talks about a Nurses Week video from Fig. She talks about Elmo: just checking in, how’s everybody doing? There’s a post from Willy Chavarria, who’s a fashion designer, leaning into the sort of joy that people are experiencing. I think you see this a lot with content creators in particular, talking about the beautiful things in their life. Like, the aesthetics of happiness feel really engaging.

Warzel: And that speaks a little, though, to the tension here, right? I know a lot of people who—I know somebody, a friend, just happens to be, they planned their one vacation for the year, you know, a year ago. And it just happens to be during this time in American life that is just awful and very chaotic, and feels just like, you know, something has shifted and changed. And there’s this, I think everyone makes all these different calculations of Do I post a photo of the beach? Like, Am I a bad person if I’m doing this? And I think it’s interesting to think of it from this lens of, you don’t want to say “The world just goes on,” right? But there is this way of thinking—that sharing and joy. And, I don’t know, there’s a way that you can create community that feels a little generative, as opposed to the constant soul-sucking nature of We are on the brink.

Litman: And not to bring everything in social media and politics back to Zohran Mamdani, but I do think this is something Mamdani did so, so well—which was that his stuff was joyful. Like, it was heavily layered with Millennial optimism and a little bit of cringe in a way that really, especially, spoke in opposition to the way that [Andrew] Cuomo or anyone else was campaigning. If you believe that better things are possible and you are inherently an optimist—because I think to participate in activism, you have to be a little bit of an optimist, even if you hate that about yourself—it is contagious. Because it’s what gives you the forward momentum to keep doing the work. So I had a friend who was doing that too; was like, I just was surfing in Costa Rica, and the world is falling apart. Can I post? I was like, Yeah, of course you can. Because the next day you’ll be back, you know, donating and marching and doing what you can. Post away.

Warzel: Or shoveling snow and doomscrolling; yeah. I wanted to ask broadly, we’ve touched on this in a lot of ways. But something that I have thought so much about—and, to be very candid, like agonized about for the last decade—is: What do we do with this window to the world that we have? I think whether it’s seeing atrocities abroad, as we have for so long, or the dysfunction here at home, there’s a feeling right now that part of just being alive today is being confronted with auto-playing videos of people being shot to death. And that’s something that I think is beyond just being hyper online. If you are in these spaces at all, you may be confronted with fresh, genuine horrors—that if you open yourself to them fully, it’s easy to shut down or to feel just so alienated, because you are going throughout your day, in the way that you are. I’m curious how you think about how we bear witness. How we think about showing and sharing these things with people because they matter to us, and how not to get demoralized in doing so.

Litman: One of my general rules of thumb for online-content consumption, especially as it relates to the news, is to only take the poison I have the antidote for.

Warzel: I love that.

Litman: I can’t always do this. But generally speaking, that means that there are huge parts of the things happening in the world that I just like, I scroll past. There’s nothing I can do about that. That doesn’t mean it’s not important. That doesn’t mean it doesn’t matter. It’s just the things that I have chosen to dedicate my life and my career and my work to, and my personal outside-of-work interests to, are not that. Now the things that I have—like what is happening in Minneapolis, like what’s happening in the United States, like what’s happening with Trump, like what’s happening in New York City, where I live—yeah, I’m diving deep. But you know what? That’s because I can do something about it. Whether that’s voting in an election, giving money, subscribing to media, thinking about all the different sort of things I could do that affect it. Pick the poison I’ve got the antidote for. And I think the key is remembering that the antidotes, to extend the metaphor a little bit, are more varied than you might think.

I was talking to someone who was telling me they felt so hopeless about what was happening in Minneapolis. They’re like, There’s nothing I can do. This person was like, I’ve worked in politics, worked as a facilitator. It’s so hard, but there’s nothing I can do. I was like, You live in D.C. There’s a D.C. mayoral election in the upcoming year. It actually is going to be really, really, really important who the D.C. mayor is, because this shit is going to come back to your city soon enough. Like, dedicate all of your energy to that. And that—while it might not feel like a direct solution to what’s happening in Minneapolis—that is a solution, and it’s a concrete thing you could do. It’s a thing you can show up in person for. It’s a thing you’ll be able to live the results of. Be expansive in what your idea of antidote is, and it allows you, I think, a little bit more possibility.

Warzel: We’re talking a lot about the personal-political media consumption, how to be online as a person. We’ve also touched, but I want to go deeper on, the idea of politicians and being online. You’ve written a lot about leadership, especially in younger generations: about ways that people can leverage these platforms to connect, to get messages out, but also to inspire people to affect actual political change. And something that I feel, and I think a lot of people feel, is that there are really just—especially when you look at, say, our politicians, they fall into two camps. They’re either hyper-online, really cocooned, troubling edgelords. Or you have people who are kind of thinking and regarding the internet on dial-up terms, right? You’ve said that we need to find elected officials who are the right amount of online. What is the right amount of online for an elected official?

Litman: So I think it’s folks who are enough of a consumer that they can be a producer without the brain rot. Like, J. D. Vance has brain rot. Trump has brain rot. Like, they’re so online that they’re unable to see that it is an ecosystem—an echo chamber that they cannot get out of. They’re sucked into it. And I think there’s actually a lot of, you know, Republican electeds, candidly, who—whether it’s Twitter or Fox News—same idea. Like, there’s not some secret source of information. They’re doing the same shit. We saw that in the photo from the Venezuela attack, or coup, I guess. Where they’re scrolling Twitter as they’re, you know, kidnapping dictators.

Warzel: Right. This is the, sort of, not Situation Room, but the Mar-a-Lago command center during the Venezuela raid, where they are conducting an operation. And Pete Hegseth has a screen behind him that is showing a Twitter feed, or an X feed.

Litman: That is too online. And then, as you say, the politicians who clearly have no idea how the internet works; don’t understand. They don’t know the language of a forward-facing video, so they can’t do it in a way that feels genuine or authentic. Which means they can’t communicate in this moment. That is 80 percent of a politician’s job—is communication. And if they can’t do it in the way that works for 2026, they can’t be a good politician right now. You’ve got some politicians, I think, like AOC [Alexandria Ocasio-Cortez], Maxwell Frost—I actually think Chris Murphy and Brian Schatz, it’s not all younger—like the ones who are online enough that they know how to use the tools. Even if they don’t actually know how to, like, edit the video themselves. While still understanding that the internet is not real life, but it is a huge part of your life for a ton of people, and it is how people understand real life. Like, it’s a mediation for how we will understand real life.

Warzel: When you see something good from a politician, like a piece of content that they’ve produced or a response, what are some of the flavors of that? Like, what is a taxonomy of a good response or a good post from an elected official? How does that work for you?

Litman: Does it feel like it credibly could have come from them? Do I buy that this is a thing? That even if, like I said, like their Zoomer digital staffer told them, Hey, this is a thing you should have, you should make, we’re gonna edit it—does it feel like it credibly could have come from them? Does it feel like they’re responding to a post that they probably could have seen? Does it look like they’re copy-pasting talking points they’ve been emailed? Which you see politicians often do, and then forget to delete the copy-paste notes on top of, which is such a flag. How much did their staff have to explain to them the format, the function, how it works? One of my favorite questions to ask politicians of any age, but especially the older ones, is “Tell me about what side of the internet you’re on.” What side of TikTok are you on? What Reddit subreddits are you in? Are you on Pinterest? Are you on Ravelry, if you’re a knitter? I don’t know; tell me about the internet that you’re on. And the ones who don’t even understand the question are the ones who are not well suited for this moment.

Warzel: It’s interesting. I once asked—I had an opportunity via The New York Times editorial board to interview Bernie Sanders, and my job was to ask him a tech question. And I was just like, Show me some of the apps on your phone. And he just yelled at me and said he doesn’t have any. And yet, somehow that works, right? Like somehow, there are some people, because he is authentic to that, like Hey man, like, miss me with all of this.

Litman: But you can tell the ones who are actually really good at it don’t do meme-y stuff. Like, they’re not making memes. They’re being, like, genuine personal communicators, who maybe are using different formats and functions. But I think part of it is they so clearly know themselves and know what they believe that the question becomes, What tactics am I using to communicate it? As opposed to, Which memes am I jumping on to become trendy?

Warzel: This is what I noticed, and other people pointed this out, about Mamdani. Again, politics aside. What was so—and also even beyond the “joy” part of it aside—what I thought was so interesting from a media perspective about some of his videos, especially the first ones that he did when he was out in Queens on the street, was he was taking a very popular online form of content. Which is the “man on the street asking random people, with a microphone in their face, some kind of question.” Right? And then that getting chopped up and put online. Like, that’s how we found Hawk Tuah Girl, who became like the other side of it. But he was just asking these questions, about like, Why did you vote for Donald Trump in 2024, right? And having that man-on-the-street thing. It’s that fluency with a type of content that people are familiar with, but it’s also a type of content that you can do in your own way that is not going to, or that is going to then spread naturally, right? Because the internet is primed to take that type of content and just get it out to a lot of different people. I thought that was a really great example of pairing form, regardless of even the politics, with something that is actually gonna spread.

Litman: Well, and that’s because he’s the right amount of online. Like, that’s a dude that has an FYP that is carefully calculated to his interests. And he’s talked about this. And I think that is telling of a guy who, at the very least, knows the language. Like, he can speak enough of it that he can get by. And I think that is true for so many of the politicians who are really effective in this moment—is they’re good enough. They’ve done enough days on Duolingo to get by.

Warzel: They’re not being guilted by the Duolingo owl logo. That’s very important, I think, in general. So if you could give advice—and I’m sure you are—but if you could give advice to the Democrats right now. And that’s even just like in choosing candidates, like how important is that part of it, like the online strategy in assessing overall whether a candidate’s gonna be good. Obviously you don’t want a candidate who is just hyper online and great at it and, you know, not good at the realpoliticking or whatever. But how do you—how should they judge that? That category “being online”?

Litman: I don’t want to say it’s everything. But I think it is a bigger part than most people would like to admit, because I think, one, it’s how people get information. But two—as we learned with Joe Biden—if you cannot effectively sell the good stuff you’re doing, it doesn’t matter how good the stuff is. Like, you’ve got to be able to continually communicate in all the ways people get information, in all of the platforms. And part of that is being sufficiently online.

But it’s also being a normal person who can show up in nonpolitical spaces and still be a normal person. I think a lot about how many members of Congress could credibly go on basically any nonpolitical podcast—chat show, influencer—and not sound like a robot. Maybe a dozen? If you’re being generous. Which doesn’t mean they’re bad people or bad politicians or bad legislators. It just means that the skillset for what you need to be an effective leader in 2026 is not the same skillset that you needed in 1990 or 2000 or even 2010, when a lot of these folks got elected. And I think it’s the same thing that you’re seeing CEOs struggle with, too: where if you are too online as a CEO, you will suffer. But if you’re not online enough, you’re not gonna know how to perpetuate the brand; you’re not going to know how to sell the work. You’re not going to know even how to influence internal communication, which is driven off of external communication, in many ways. It really matters.

Warzel: Something that I really appreciate about your writing on this entire subject is always that it draws back to IRL, right? It’s always focused on the idea of actual civic engagement, actual community. The stuff that we do online driving change in the real world. You’ve written about things —you’ve written a long list on one of your Substacks of things that people can do. Some of it is political; some of it is civic engagement and relationship building. Can you tell me some of the things that you have found most helpful in just building community, and giving people just a broader sense of agency?

Litman: Go to a city-council or school-board or library-board meeting. Go to a state legislative hearing. Go to a neighborhood cleanup. Stock your neighborhood fridge, if it has a free fridge. Participate in mutual aid. Support a local artist. Buy something from a local artist. Share the thing that you bought online, so that more people can buy things from that local artist. Subscribe to local media. Bring a dish to someone who just had a baby or lost a loved one, or just because. Start a neighborhood group chat. Sustain the neighborhood group chat and make sure it’s not that annoying. My favorite one—and I’ve written about this a lot, it’s sort of like the other thing that I’ve been clinging to this year—

Warzel: I think I was gonna ask you about this one. Is this your experiment?

Litman: Yeah.

Warzel: Tell the people.

Litman: In 2025, and I’ve written about this a bunch and talked about it lately, my husband and I—in order to combat our new-parent isolation—decided to host people for dinner every Saturday in 2025. Every Saturday we were in town, which since we started the year with a 2-year-old and a two-month-old, was every Saturday, because we didn’t travel with them. And we ended up succeeding. We had 52 dinners. More than 100-some-odd people came to our home over the year, 40-some-odd kids. And it was hard in many cases, but so magical. And as I think about, like, Okay, what happens when ICE comes to New York—which they will, they are, I see it in the neighborhood chats already. Like, When they send tanks rolling down Flatbush Avenue in Brooklyn—not a crazy thing to say might happen at some point in the next couple years—who are the people that I’m gonna text? Who are the people I’m gonna turn to? It’s the folks who’ve come over to my home for dinner, and their friends, and their networks. I described it as the most political thing I did in 2025. And I am a political operative who voted in two different elections in 2025—thank you, New York City—and give money and read and engage. Having people over for dinner was the most political thing I did. And as I read about what’s happening in Minneapolis, I feel even more affirmed in that. Like, building strong relationships with the people around you is what will get us through this.

Warzel: What did you learn about being social in 2025, 2026? This era where people do feel really isolated a lot? There is a quote-unquote “loneliness epidemic,” however you want to describe it. What did you learn about being social from having 52 dinners at your house?

Litman: One, that it really surprised people to be invited over. People were shocked that we were inviting them into our home. Two, that the tone that we set really mattered. How casual could we make it? I’d often text people morning of to be like, Hey, just FYI, one of my children is not wearing pants and I’m still in my pajamas. See you later. Like, being really aggressive about making it clear—this is not fancy; this is as real as it can get. Because it’s all we can do at this stage in our lives. But come into our home anyway.

I would say some of the questions that I’ve gotten since I’ve written about this and posted about it—because always be posting—really have stunned me. People would ask, like, So what did you talk about? Books, movies, vacations, our kids, the dog, all kinds of things. But I do think that social muscle—and I don’t mean to embarrass anyone by saying that, it’s a real question—that social muscle has so clearly atrophied, and you so deeply overthink it to the point that you have forgotten that actually you can just let go. Not to Mel Robbins you, but you can just let it go.

Warzel: You know, I love—I’m probably butchering it, but I believe it’s like a [Kurt] Vonnegut quote. I think it’s like, The humans were put on this earth to to fart around is the quote, right? And it’s like, I think about that. I thought of it when I read your post, and I think about it sometimes when I am out with my community, engaging in whatever. And a lot of that time is like … it’s not structured, right? It’s not like, “Here’s the list of what we’re gonna talk about today, and the eight political actions we’re gonna take, and the 12 things.” It’s like, I almost sometimes can’t remember what we talked about, right? But it’s this, leave with this feeling of, like, That was so great and unproductive, because it was unproductive. It’s like a wonderful thing. And it feels like it’s something that we just like, we need to have. Just that unstructured hang.

Litman: It’s really hard to be on your phone when it’s just four or six people in your house hanging out and having dinner, drinking wine, whatever it might be. Which I think is also big part of this; it’s two or three hours in which I am not on the internet. And I wish that was more of my life. I wish I was better about not being on my phone. But like, especially when people are over, it is so valuable to unplug. So valuable.

Warzel: Well, I think we will let people unplug by wrapping this. Amanda, thank you.

Litman: Excellent segue.

Warzel: I’m trying. This is still new to me here. We’re working on it. Amanda, this is lovely. Thank you so much for helping people try to understand how to use these tools to feel like a person in the world, instead of being tossed by the algorithmic winds. Thank you so much.

Litman: Well, it’s my pleasure. I would just say—I know it feels bad. We’re winning. We’re going to win. I think that’s the thing I’ve taken away from the last couple days of the internet. It’s like—they’re not popular. What they’re doing is not popular. They’re getting cat-bongo Reddit subthreads to vocally disagree with them. They are not popular. This is terrible, but we will win. And I find that to be really comforting.

Warzel: Amanda, thank you so much. Really appreciate it.

Litman: Thanks.

[Music]

Warzel: That’s it for us here. Thank you again to my guest, Amanda Litman. If you liked what you saw here, new episodes of Galaxy Brain drop every Friday. You can subscribe to The Atlantic’s YouTube channel or to Apple or Spotify or wherever it is that you get your podcasts.

And if you want to support the work that I am doing, and the work that all of my colleagues at The Atlantic are doing, you can subscribe to the publication at TheAtlantic.com/Listener.

Thanks so much. See you on the internet.

[Music]

Believe Your Eyes

2026-01-25T13:49:06-05:00

Chances are, you’ve seen Richard Tsong-Taatarii’s photo. Taken Wednesday in Minneapolis, it shows an unidentifiable protester face down on the ground; two Border Patrol agents are on top of him, holding him there, while a third unloads pepper spray into his face from just inches away.

Tomorrow’s front page of the Minnesota Star Tribune: Jan. 23, 2026

[image or embed]
— Minnesota Star Tribune (@startribune.com) January 22, 2026 at 6:19 PM

The photo ran on the front page of The Minnesota Star Tribune on Friday and already feels like a defining image of the long ICE incursion in Minneapolis—a powerful illustration of how the agency has acted, in broad daylight, with excessive force and impunity.

This is just one example of federal agents terrorizing people in the city. There’s also the photo of a 5-year-old boy being detained outside his home. There’s the video of an agent chasing a teenager through the snow on a residential street as the boy yells “I’m legal” in Spanish. And yesterday, the world saw footage of Alex Pretti, a nurse who worked in the ICU of a Veterans Affairs hospital, as agents pepper-sprayed him, knocked him down, appeared to remove a legally permitted gun from his person, and fired at least 10 bullets into his prone body. (Border Patrol Commander Greg Bovino has claimed that the agents were the “real victims” and suggested that they were acting in self-defense.)

[Jonathan Rauch: Yes, it’s fascism]

There are still unknowns in this case, as there are in many of the chaotic moments that have come out of ICE’s recent surge in Minneapolis. But there are basic facts: Pretti was helping a woman when agents swarmed him. He did not have a gun in his hand. In the past 18 days, agents have used excessive force against numerous people in Minneapolis and killed two of them—first Renee Good, now Pretti. We know about this violence—we can see it ourselves from numerous angles—largely because of video and photographic evidence taken by everyday citizens, many of whom have purposefully set out to make sure that they are recording what is happening for the world to see.

Tsong-Taatarii credits these volunteers, many of whom are trying to protect their neighbors, for his now-famous shot. On Wednesday, Tsong-Taatarii had been following Bovino but realized that was “a wild goose chase,” and was alerted by a group to an escalating situation in South Minneapolis. He drove over and got the photo. When I spoke with him on Friday, he told me that protest observers have set up Signal group chats, which help track agents’ movements across the city. He uses the chats to make sure he can be in the right place at the right time to document what is happening.

If it still is not clear why this matters, compare the documented reality with the Trump administration’s blatant propaganda. The Department of Homeland Security said in a statement yesterday that Pretti “wanted to do maximum damage and massacre law enforcement.” Similarly, White House Deputy Chief of Staff Stephen Miller called Pretti a “would-be assassin” who “tried to murder federal law enforcement.” Video footage directly contradicts these claims. It shows Pretti holding a phone in his hand, pointing it at an agent after another shoved a woman. He was shot again and again while on the ground.

The full evidence around Pretti’s death has not yet been released. If he was indeed filming agents before he was shot, there is likely first-person footage on his phone. But Pretti’s last seconds were captured from multiple angles, in sickening footage widely distributed on social media and by news organizations. It is able to be seen and dissected online precisely because of the observers who were there to document it, who watched as federal agents piled atop Pretti and who did not drop their phones when the gunshots rang out.

[Read: Minneapolis is a Second Amendment wake-up call]

The work of observers and photographers in Minneapolis right now is as dangerous as it is crucial, because ICE’s presence in Minneapolis is provoking not only physical conflict but an informational conflict. Agents themselves are pulling out their phones during altercations with protesters. According to The Washington Post, the White House has urged ICE to “produce videos for social media of immigrant arrests and confrontations to portray its push for mass deportation as critical to protecting the American way of life.” Last week, President Trump posted on Truth Social that ICE must “start talking about” the people they’re arresting in Minnesota, writing: “Show the Numbers, Names, and Faces of the violent criminals, and show them NOW.” When the footage doesn’t suit the administration, it seems to have no issue doctoring images to suit its alternate reality, as it did on Thursday. Agents had arrested an attorney who was protesting at a local church, and the White House posted a photo of this woman that was altered, presumably by AI, to make it look like she was crying.

A dark irony of our current age is that there is more video and photographic evidence than ever before, and yet propagandists can coerce or convince others to not believe what they can see with their own eyes. (See also: January 6, 2021.)

If the truth is ever to win out over propaganda, it can only do so in the face of overwhelming evidence, the collection of which has become ever more treacherous in the second year of Trump’s second presidency. But the news and footage of Pretti’s death seem to have broken through the usual informational chaos—at least to some extent. On Reddit, Instagram, and Facebook pages, the videos of Pretti’s last moments appear to have galvanized people who don’t normally engage or post about politics. And it is thanks to the bystander videos of Pretti’s killing that people are trying to hold the administration accountable. This morning, the CNN host Dana Bash referenced the footage in an interview with Bovino, asking him why Pretti was shot after being disarmed. “We’re not going to adjudicate that here on TV in one freeze frame,” Bovino replied. “It’s not a freeze frame,” Bash said. “We’re showing a video of one of your agents taking the gun away. And that happened before Pretti was shot.”

Minneapolis residents are risking their lives to document what is happening to their city. In Pretti’s case, doing so cost him everything. We should believe what we can see with our own eyes. One can only imagine what Miller and the administration might have said about the shooting and Pretti if there weren’t an abundance of footage. Thankfully, because of the observers, the world can see for itself.

ICE Is Turning Real Conflict Into Viral Content

2026-01-23T13:30:00-05:00

Subscribe here: Apple Podcasts | Spotify | YouTube

In this episode of Galaxy Brain, host Charlie Warzel speaks with the reporter Ryan Broderick about how the internet’s fragmentation of attention and facts has bled into real-world political violence in Minneapolis this month. From the viral spread of a right-wing video about day-care fraud in Minnesota to the aggressive ICE activity in the region that followed, the episode charts how online content routinely shapes government action and public perception.

Broderick, who spent days in Minneapolis after the shooting of Renee Nicole Good, describes what he saw on the ground: how protesters and law enforcement are behaving differently this time around, especially with regard to filming and digital organizing. The conversation explores a novel and concerning feedback loop where what happens online spurs real-world interventions, which then generate more content for audiences elsewhere, compounding division and uncertainty about what’s true.

The following is a transcript of the episode:

Charlie Warzel: Going back to Renee Good, the idea that there was an ICE agent that was filming while involved in this life-or-death—you know, supposedly for him—situation, right? You’re claiming that, but at the same time you’re using your phone to document this.

Ryan Broderick: Yeah, I’ve never had a law-enforcement agent pull up their assumedly personal smartphone and film me. I’ve never seen that—to have them just, like, have a gun in one hand and a phone in the other blew my mind. And I just have to wonder, Where is that content going? Where are those photos and videos going?

Charlie Warzel: I’m Charlie Warzel. Welcome to Galaxy Brain. Around 2015, I started to do a lot of reporting on the ways that the internet was causing a societal rupture. It’s possible, I think, that for Americans—and maybe for everybody—that there’s never been a true shared reality. That we have always been living, in some respect, in filter bubbles of our own making, even before the internet.

But as the world really came online, and as our politics and our culture were uploaded to these social networks that are essentially built for viral advertising, the shreds of our monoculture were consumed by a constellation of content, much of it algorithmically fed to us, and that something changed in that moment. I wrote back in 2017 that it felt like that was the first year—in the first year of the [Donald] Trump presidency—where it just became obvious that Americans were effectively living in two universes, and each one was propped up by their own information ecosystems.

Around that time, too, I was also writing a lot about the ways that technology would continue to blur and make it harder to understand what was real and what wasn’t real. In 2018, I wrote this piece where I spoke to a number of researchers, including a man named Aviv Ovadya. And he specifically warned of this infocalypse, is what he called it—an information apocalypse—basically a future where AI tools and craven hyperpartisan actors on this algorithmic internet that’s completely fragmented and fractured and tech-powered would allow anyone to cast aspersions on any event or anything or any fact. People could just fake anything. And they argued that would eventually create this realm where, to borrow a phrase from the researcher Peter Pomerantsev, nothing is true, and everything is possible.

People, they said, would be radicalized. They’d be overwhelmed. They’d be hopelessly divided, and they’d be sad. But mostly it would get to the state where when you can’t figure out what’s true—when everything is just this information war—that eventually you just start to tune out. And this was all pre-pandemic. It felt very plausible, but the generative AI tools weren’t out yet. They were pretty crude. They were pretty bad.

So there was this feeling that it was plausible, but it still felt so futuristic.

But today, it kind of feels like we’re living in that future. Just in January, just this month, it feels to me—as someone who’s covered this for a long time—that we are living in a pretty straightforward depiction of what the worst-case, post-truth, information-apocalypse scenario was, of the people who were warning about it in the late 2010s. I mean, if you just think about this month, you have X and Grok and this major platform generating and distributing this AI nonconsensual sexual-abuse material. And having it viralized and weaponized to intimidate and harass minors and women. You have a terminally online government. And you have the officials marshaling the resources of the military in service of propaganda to perform for this imagined audience online. And that’s all happened with the U.S.’s military invasion and capture of [Nicolás] Maduro and all the content that has been posted alongside it. You have a YouTuber named Nick Shirley, who at the end of last year posted this video online alleging daycare fraud in Minnesota involving the state’s Somali population. That video exploded online and eventually led to a chain of events that led to the Trump administration deploying a higher level of enforcement of ICE en masse to the Twin Cities. This, of course, culminated in an increased amount of raids and ICE agents in the street in Minneapolis. And, ultimately, in an ICE agent shooting and killing Renee Good in her car. It was captured on video. That video quickly spread online, with people analyzing it from multiple angles There have been tons of protests in the ensuing weeks. So much of that has been captured online. Protesters filming ICE; ICE agents filming protesters. Incredible amounts of content on social networks of people being abducted or arrested or having tear gas deployed on them.

It is this situation where it feels like there’s this battle being fought through everyone’s phone. It feels a bit like a hinge point, and it feels like there is a—despite all of the political issues happening here—that there is a technological issue, that this is all a bit of a culmination of things that I’ve been following for a really long time. Somebody else who’s been following that in some cases alongside me is Ryan Broderick. Ryan Broderick writes the Garbage Day newsletter and is the host of the Panic World podcast. Ryan and I worked together at BuzzFeed during the 2010s for quite a while, covering the ways that the internet changes how we behave politically, and the way that it can impact some of these political movements. Ryan was someone whom BuzzFeed sent to cover all kinds of online and offline protests around the world. He’s been to 22 countries, reported from six continents. And he’s been on the ground for close to a dozen referendums and elections. Ryan recently came back from observing all the protesting in Minneapolis, and he joins me today to talk about what happened in Minnesota, what is happening, and the ways in which all of this can be linked and not linked to an extremely online society. He joins me now.

All right, Ryan Broderick, welcome to Galaxy Brain.

Ryan Broderick: Thank you for having me. I’m so excited. Congrats on becoming a podcaster. I’m really sorry that happened to you.

Warzel: I know. It’s everything I ever wanted—to just be mocked on YouTube and across multiplatforms. It’s lovely.

Broderick: That’s right. It’s about discovering new audiences that hate you. That’s what modern life is.

Warzel: That’s what we’re doing. Man. Well, I appreciate it. I want to talk to you about your reporting. You went to Minneapolis just a few days after an ICE agent killed Renee Good, and you were there for protests around the city and around the federal building. I want to start with—just tell me what it was like to be on the ground in Minneapolis for those few days.

Broderick: It’s very surreal. I’ve definitely come back with a profoundly different take on the seriousness, I would say, of the Trump administration. I think you and I have been having essentially one conversation for a decade. Which is like: How seriously should we be taking this stuff? Is this just like internet chatter, or is this a real existential threat to American democracy? And now that I’ve kind of seen the whole hyper object—you know, from internet content down to what is effectively an occupation of an American city—I think we need to take it very seriously. I’m pretty freaked out.

Warzel: Yeah. So give me a sense of, because you were kind of there in the beginning of this. I mean, things are still going on. There’s still people protesting. This is not a finished movement in terms of the raids in the Twin Cities. But since you were there in the beginning, tell me how you saw it evolve, in terms of both the protesters and ICE’s participation there.

Broderick: Yeah. So, we saw the news come across social that Renee Nicole Good had been killed in Minneapolis. And my producer, Grant, and I—we work on our show Panic World together—so we started talking to our partners on our podcast, which is Courier News. And I was like: “Look, I think this is big. We should do this.” And so we went. Basically, we were there 24 hours after she was killed, and we spent the weekend there. And, you know, if we had had the money and the resources, I’d still be there.

I think it is still unfolding and developing. But the point of our trip was not to—I keep stressing this in everywhere I talk about this—like, our point was not to re-report what people in Minnesota are already covering. They have a great local news scene. But there are specific questions that I think people like you and I ask of events like this, that local news doesn’t have the bandwidth or even the interest in asking, or know. And so it was really useful for me to just follow the action for four or five days and see how this community is using technology, how they’re responding to technology, how the internet is shaping the events on the ground. And I think we did that, but yeah, I think this is going to become just sort of like one of those long political quagmires that happens during Trump administration. Like, it’s just going to drag out.

Warzel: So give me a sense of the level of online-ness of these protests. Give me a sense of how the tech is shaping it. I think for the first, part of the question is: A lot of modern protest movements are very online organized, via social networks and different apps and things like that. This one seems to be, based off of the reporting I saw a little bit. You read that a little bit differently.

Broderick: Yeah. I mean, so—I was trying to think back of what was the last protest that I went to. And I went to a few during 2020. But the last sort of movement like this that I saw close was the Hong Kong democracy movement in 2019. And I was there for what would eventually be—I think no one knew at the time—but it was the last time they protested before China really cracked down. And that was totally online. Like, I watched a completely silent crowd moving in perfect sort of organization using Telegram. Which I was really like, freaky. lLike these Gen Z kids had figured out how to use these messaging apps in a way that I’d never seen before. Minnesota is not like There’s a hashtag, and we’re all spray-painting it around the city, and we’re wearing Guy Fawkes masks and shit. No, none of that. It is people who are, if they’re using apps, they’re using Signal. They’re using a walkie-talkie app; I think there’s a couple that they’re using to organize ICE monitoring. And they’re finding out about protests and demonstrations by just going to them. Like, we interviewed countless people who are like, I was on my way back from work, and I saw this, and I’m mad. So I came by. And we talked to people who’ve never been to a protest before in their lives. Like, you know, we’re talking like middle-aged people who are like, I don’t have any real political thoughts, and this is insane. And I’m going to come out, and I’m going to protest. So it was very different. And to compare that to the other side, which is extremely online, was a total inversion of everything I’ve seen in my career. Like, it really stopped me in my tracks.

Warzel: Let’s talk a little about that. Because something striking that I saw in your reporting, just like coming across on social media as you were like sharing in the moment, was—

Broderick: You didn’t read the post?

Warzel: Well, I always read the posts, but in the moment.

Broderick: No, you just read the Bluesky post. See, I know. I get it. Yeah.

Warzel: I mean, you think I’m gonna pay for a subscription? I mean, come on.

Broderick: No, I wouldn’t. Yeah. Of course.

Warzel: Something that I saw, though, was you were videoing a couple of what seemed to be content creators, like stepping out of ICE vehicles. Or at least government or unmarked government-looking vehicles. Filming, and then like popping back and not addressing themselves. What did you see in terms of creators, influencers who were sort of either deployed by or administration adjacent? Or supported by the administration, you know, in some facets?

Broderick: So the specific incident you’re talking about, with the guy who jumps out of the ICE vehicle. Basically, this woman gets arrested by ICE because she punches their window as they drive by. Like, didn’t damage the window or whatever, if anyone cares. But they jump out of the cars, they pull up, and they just tackle her to the ground. And then right behind them—coming out of the same car—was a guy in plainclothes, like a nice suit. you know, like sunglasses. And he’s smiling, and he’s filming the whole thing.

And I am screaming at him, like, Who are you? Why aren’t you in uniform? Who are you with? Where’s that video going? I later find out that it’s a Fox News correspondent that was on a ride along with ICE. But there was also just like a ton of other right-wing networks that are around. Like, the Daily Wire was filming people. And OAN and NewsNation and all the baddies. There were also a group of YouTubers, livestreamers, content creators that were running around the city, harassing people. And they had a much less official connection to ICE. But obviously whenever they would rile up the crowd, the ICE agents would come out and protect them from the crowd.

And that culminated in last weekend with the leader of that group: Jake Lang, a January 6 insurrectionist. He tried to burn a Quran on the steps of the Minneapolis City Hall. And got his ass beat. And so it’s—that, to me, is much more connected to what we’ve already seen in Trump administrations. Like, the roving gangs of content creators that are just, you know, trying to antagonize leftists and make them look violent on stream. But the fact that ICE was so brutal to the protesters while also basically running defense for these content creators … they weren’t trying to hide it, in the way that you know, police forces were in the 2020 protests.

Warzel: Well, let’s talk a little bit about what you saw from ICE agents. Because there is this element, and I want to get into it in a second, of this content-creation spectacle. But in terms of you saying you came back from this, and it’s very obviously understandable why you would feel this way. Like, I’m freaked out about this. And the seriousness of what’s happening there and what the administration’s doing. But can you just describe for me—and for people, you know, who may not be mainlining this stuff on their phones—what you were seeing from ICE that was so disturbing, or the way that they were treating protesters.

Broderick: They don’t even seem to understand why they’re trying to control the crowd. They don’t, really. I’ve dealt with, you know, antagonistic police forces in protests before. And there’s a rhythm to it. There’s sort of like an assumption that like, If you do this, they won’t really hassle you. And, you know, we can get into the weeds of like how that breaks down as protests become more extreme.

But, like, there’s protocols. ICE does not have that. They don’t really care. They will scream at you. They’ll yell at you. They’ll kind of break kayfabe. They are also filming you. They are, like, very intent on filming you with their cell phones. And with telescopic lenses, that they have guys parked behind them in cars, filming through the window.

Warzel: And that feels very different, right?

Broderick: I’ve never seen that. I’ve never seen that.

Warzel: That’s not something I’ve … even going back to Renee Good, the idea that there was an ICE agent that was filming while involved in this, you know, life-or-death—supposedly for him—situation, right? You’re claiming that, but at the same time you’re using your phone to document this. That feels very unique. That feels like a very different form of, you know, enforcement tactics.

Broderick: Yeah, I’ve never seen it. I’ve never had a law-enforcement agent pull up their assumedly personal smartphone and film me. Like, I’ve never seen that. And I mean, obviously, like there are agencies that do surveillance on protest movements, and it happens. But to have them just, like, have a gun in one hand and a phone in the other blew my mind. And I just have to wonder, Where is that content going? Where are those photos and videos going?

Warzel: Yeah. Anyway.

Broderick: Yeah, And so, like the fact that these guys are just taking photos of people and either uploading them to some kind of database or sharing them with each other, like, my mind sort of goes to: Okay, then what? So these guys, let’s say they have dozens of photos and videos of random people in Minneapolis on their phones. Like, are they sending them to each other? Being like, Keep an eye out for this person, and like beat their ass if you see them? I don’t know. We will eventually, at some kind of trial, perhaps find out what these guys are doing on back channel. But yeah. It sort of adds to the idea that these guys are like, I don’t know, content creators first and a Gestapo second. Or something. It’s very surreal, very dystopian.

Warzel: Yeah, and the nonprofessionalization of all of it, right? There is a piece, I believe it was last week, could have been the week before, where a reporter, writing for Slate, applied to work for ICE. Had a military background, but certainly was like somebody who had been hypercritical of the Trump administration and immigration policy, enforcement policies, and ICE in general. And still got hired. The idea of being, like, the threshold is so low. Did it add to the anxiety knowing that you’re dealing with a sort of improvised force? Not only that the guys are out there filming—it seems like they wanna be creating some content—but also just this idea that, again, the standards of training might be totally different. Did that add to the sense of unease, in terms of the protesters and the people who are around?

Broderick: Yeah. I mean, you can’t really predict what they’re going to do. I remember on the Saturday we were there, there’s this massive protest in Minneapolis. Thousands, 5,000 people. Many of them immigrants. Like, many of them with immigrant-rights groups and stuff. And, you know, we’re going through these little side streets in Minneapolis. And I just kept thinking, like, What is stopping Trump from saying on Truth Social, “Hey, get them”?

And Greg Bovino and everybody, like, cordoning off two sides of the street and just kettling everybody? You know, that can happen. Like, the sort of general unease of the whole thing is that you’re not dealing with an accountable, a rational, an identifiable law-enforcement group. You’re dealing with guys in masks that are filming you that don’t really have any kind of protocol. And don’t really seem to care about anything other than quotas.

I read that they’re supposed to be getting, like, 3,000 arrests a day. And the other thing is like: When they do detain you, in a normal situation, a reporter gets apprehended at a protest. You might get thrown in jail for like a few hours; maybe the weekend. You might miss filing your story. And then, you go home. With the ICE detention facilities … like, they might just put you somewhere.

And so, if you’re a reporter, and you’re trying to cover this, they’ve already made filming their activities a possible crime. According to what is NPSM-7. If you film their activity—

Warzel: Yes.

Broderick: —they can charge you with terrorism. They can charge you with obstruction.

And there’s really no oversight for what they’re doing. So the risks are just a lot higher than any sort of protest movement I’ve ever seen. Yeah.

Warzel: And I believe, what is it? It’s an NSPM-7, right? National Security Presidential Memorandum 7.

Broderick: NSPM-7, yeah.

Warzel: Is basically an executive order that is meant to dismantle left-wing terrorism and basically allowing the government to qualify people as “domestic terrorists” or “domestic terrorist organizations,” which allows for different standard of prosecution. And yeah, so that’s adding to all that confusion and chaos. So you saw this. You wrote, “We are all content for ICE.” It’s very clear that the Department of Homeland Security has basically transformed immigration arrests into this visual propaganda online. It’s very clear that there is a strategy here. There is a lot of pressure from the administration to have ICE really active in terms of a media campaign. What purpose do you think it is serving them to do that?

Broderick: I think everyone is trying to answer this question. And this might be an unfulfilling answer, but I think there’s a couple things happening simultaneously. I think, one, I think the Trump administration is full of genuinely xenophobic ideologues that want blood-and-soil nationalism. And they love this stuff. I think that Trump is a creature of the media, and he sort of understands that everything has to have a media component to it. And we’ve seen that … like, I don’t think that he’s playing four-dimensional chess. But I think he understands that everything he does has to have some kind of news cycle attached to feel like it matters for him.

Warzel: Well, I’ll just say: On January 20, which is Tuesday of this week, Donald Trump posts on Truth Social, like, We need more photos of people who we’re arresting in Minneapolis. We need more documentation of just how bad they are. Like it’s very clear that while he’s not playing four-dimensional chess that we know of, he is, as you said, just hyper-aware—

Broderick: Yes.

Warzel: —of the idea that this needs to be fed. People need to have this raw meat of “Our nation is under siege” in order to continue to support and to justify this.

Broderick: Right.

Warzel: Yeah, anyway.

Broderick: And I think the last part of this is: I think it is safe to assume that the Trump administration is inside of their own filter bubble. I think this is happening across culture en mass. And I think, based on the reports we have seen coming out of sort of the inner circle—you know, if you want to believe them—they were surprised that people were pissed about this. And I don’t think Trump is actually particularly online. I’ve never thought that. I’m like, I don’t think he knows what 4chan is—but I do think that he has surrounded himself with podcasters and posters. And I think that those people are super tapped into X.

And so, we haven’t actually talked about this dimension of what’s happening in Minneapolis yet, but I think it’s important context. So ICE has been in the state for months. All of this ratchets up after the YouTube documentary by the right-wing influencer Nick Shirley. Nick Shirley’s documentary alleges fraud at day cares in Minnesota. All of it had been reported before. It was even being investigated by the FBI already. Like, we’ve known about this. But the right-wingers who had never heard of this—because they don’t read real newspapers, and don’t have accurate information pipelines—were incensed.

So that, to me, points in the direction of: These people genuinely do not know things. And you can say it’s by design. You can say it’s because they’re all stupid. You can say it’s because they’re addicted to the internet. Could be a combination of the three. But that does inform their behavior. And we’ve seen this all year, all of last year. That they are—the administration is doing things because they see them online.

Warzel: Right. And also, think, too: There’s the other side of this. And you wrote about this. I can’t believe this was also in January, when the United States invaded Venezuela. But, you know, the idea of a content-first administration. But something you wrote—I’m going to quote you to yourself here. Which is always fun, I know. You said, “Politics—and political violence—is now something performed, first and foremost, for an online audience. It almost doesn’t matter what happens IRL if it makes noise online.”

And I think that, you know, that it’s not just what you’re saying—which is that there is this filter bubble. There’s this idea that this is acceptable, right? Like, ltheir whatever, Overton window or whatever you want to call it. Like: This is fine. We can do this. People won’t be mad if we act in this particular way. But there’s also the idea of that imagined audience, that maybe someone sees on X or some other place or whatever. But it’s this idea of almost one-upsmanship. It’s like, you do this video of what’s happening in Minneapolis, supposedly with day cares and Somalian refugees. And we’re going to basically perform fan service by sending a paramilitary force into that town. It seems like that level of the performance is really, really important here.

Broderick: Yeah. So once again—for people who are not like mainlining this stuff all day—and they’re sort of thinking like, okay, Trump too. It’s a little more in your face. It’s got like a fresh new attitude, but it’s like—basically it’s the same idea. And there’s, like, a core loop has changed. Trump: One, he wakes up in the morning; he watches Fox & Friends. He performs the day. He then finds out the next morning how Fox & Friends has, like, digested what he’s done. And he just keeps going. Right. So everything’s kind of performed for this TV channel that is then sort of synthesizing what he’s doing and then influencing what he’s doing.

So the difference, though, between that loop and this one is that the internet is inherently a two-way street. So it’s not just what was sort of downstream in the first administration—like, the internet lights up because Trump does something, because Trump saw it on TV. And then TV lights up, and then TV—like, the role of TV in that equation kind of organizes things in a way. Was like a rhythm, too; there’s a much more sort of reliable rhythm to it.

This is way more chaotic. Because, like, Nick Shirley: He’s making a video based off of what he’s already seeing on the internet. It then goes back into the internet, which then influences what’s happening in reality. Which then influences what’s happening on the internet.

And I was sort of trying to figure out: Okay, when did this this new loop start in earnest? And I think I have it, which is the playbook. The test drive for all of this was Springfield, Ohio, and the conspiracy theory about Haitian immigrants eating cats and dogs in the park.

That whole thing, though, is a perfect illustration of what is now happening everywhere all at once. And it’s happened, and I’ve even seen examples of it happening in Venezuela, happening in Greenland. Like, it’s this roving, kind of, internet-content machine that just sort of lands where you live and then turns everything that’s happening to you into this completely inexplicable content cycle. That, you know, for people on the ground, makes no sense—because it’s not really meant for them. And it comes with conspiracy theories and viral videos and racist memes and AI imagery and paramilitary violence. Like, it’s all kind of part of the same thing.

Warzel: It’s really interesting to think of it as almost like a storm system, that develops outside and just comes to your town. I have had trouble trying to get a sense of who this is helping; who this is hurting. Right? ’Cause on one hand, you have all of this documented evidence of what is essentially a paramilitary force drawing guns, you know, in public places on people. Deploying tear gas and pepper balls and all kinds of, you know, chemical sprays on citizens. Arresting people on whims, intimidating people, mocking them. And there’s obviously, of course, the video of an ICE agent shooting a woman in broad daylight.

And that all adds up, I think, in the eyes of people watching online or wherever. That is doing something right to people, and how they view the country. Their country. And then at the same time, these protests can also—as we’re noting—give the administration a little bit of what it wants, right? It gives them content in order to depict these cities as war zones.

And do you get a sense of if this is, like, serving ICE’s propaganda needs effectively right now? Or if this is in some ways just information that is radicalizing people? To get them to, you know, wake up to the political reality in the United States. Where do you see that balance? Or like, where that’s coming out at the moment?

Broderick: That’s like the million-dollar question, I think. I mean—

Warzel: Right. That’s all we ask here on Galaxy Brain, is the million-dollar questions.

Broderick: Thank you for asking me the most complicated question of our time. I think there’s a couple of ways to think about it. And they are—this is always my cop-out, but I do think in this case, if American monoculture ever did truly exist, we like to imagine it did, you know, pre-Facebook. But maybe it didn’t. Maybe, like, the TV and movies made it feel like it existed, and we could all sort of believe in that. But let’s assume things are different. I think there are vastly different realities happening simultaneously in America. So there are people who are just not going outside, and they are consuming all of this online. And they are being radicalized either for or against it. And that’s, like, one camp.

And then there are the people that it’s happening to directly. The ones I talked to are just the most average Americans you could possibly imagine. And they’re like, This is insanity. And do I have to, like, go to the federal building and blow whistles every day because these people won’t leave me alone? And I’ve met those people, and they are just like the most normal people you can imagine. And they are incensed. They’re furious.

And then I think that there are a lot of Americans who just really do not believe that this could happen to them. And those realities, like, don’t really line up anymore. Because we’re not all looking at the same feed of information. We’re not all looking at the same screen. And it adds to the chaos. It adds to the unpredictability of all this. Because, you know, there are a lot of, like, non-white immigrant Groypers that sit online all day talking about how they love Nick Fuentes and think the Holocaust didn’t happen. What happens when ICE breaks down one of their doors? Right? Like, there are a lot of these keyboard-warrior types who do not imagine this could ever happen to their town. And if it does happen in their town, it’s going to happen to the people who deserve it in their town. And yet, time and time again, for the last year, we’ve seen stories come out, being like: I didn’t think they’d take the good immigrant in my town, that I liked. You know?

And so there are—I don’t think the average American really is prepared for the severity of what’s happened. Like, of when it happens to them. And it is in, I think, the Trump administration’s best interest to flood the zone with images of this stuff. Because it does make it feel abstract. The average person watches and cares until it no longer looks like it could happen to you. And I think that does serve Trump’s interests. If Minneapolis looks like a completely unrecognizable war zone, that’s great. It just means that the person watching goes, Well, that doesn’t look like my backyard. So I should be fine.

Warzel: Mm-hmm. Right. And I was just really struck in your guys’ reporting for Panic World, which came out in a video-podcast form as well—which, people should go watch that. You know, some of the people you interviewed were like—and this has been documented elsewhere as well—the most Minnesota Nice. You know, like just sort of, Yeah, you know, I just came from my shift at school. Or whatever. It’s like when you see that and you feel that there is this way of just being. Like—I think it brings the sort of, it’s a great contrasted image with the paramilitary force. It’s like, anyone who’s drawing guns on this person? Like, we’ve totally lost the plot. It’s not that it’s justified when it happens to people who don’t, you know, maybe look like you. But just this idea of … it’s a strong visual image for people who are watching at home, is what I mean.

Broderick: I think it’s also a self-perpetuating machine. Because—forget who made this point. Maybe it was Wired. I think it was Wired last week, this week. They’re all blending together. But they were basically like: There’s no reason for the Proud Boys to exist anymore, because ICE exists. And so if you’re running … I mean, Trumpism doesn’t really have an ideological center, but if you had to try to define it, you’d say, Okay, it’s like grievance-based. It’s like the Trump supporters hate other people for various reasons, and we’re gonna make a big tent where all your different grievances kind of live in semi-harmony together. And we can hurt others.

And then if you open up a paramilitary force, that requires what? 4Forty-seven days of training to join, and they’re not even gonna like check if you’re a Slate reporter or not. Like, you can just go and hurt other people. And so, in a sense, almost, it almost makes logical sense that this would be the next step for Trumpism. Because you can’t sustain that, like, othering of Americans long-term, without some way to get people to sign up and do it. So it’s almost like we have to create this paramilitary force, and we have to start paying people to do this. Because if we don’t, that energy could dissipate. Or they might just find out that maybe we’re not also different after all.

Warzel: This was something that I was thinking about in different way. And it is a bit of a tangent. But I’ve—as you have been writing about, you know, MAGA influencers, and stuff like screeching online about “civil war” since like 2017 or whatever, right? Like, people who are just like, This is going to lead to some kind of, you know, massive crack-up type thing. T

Then so many people that I’ve talked to who are actually, you know, experts in this type of stuff and in conflicts are like: “You know, part of the reason why terms like civil war are like so unhelpful is because it’s incredibly difficult to imagine in certain places, especially a country as big as the United States, giving structure to a conflict like that.” This idea that’s like, so much of it would just be, you know, weird infighting. Or, you know, acts of almost seemingly random terrorism or whatnot. Right? And the thing that I have found so interesting and scary about the way that the Trump administration is not only using ICE, but the way that they are marketing it, right? This idea of all of this propaganda that’s very clearly aimed at white nationalists. “Defend the homeland” type of stuff.

The reason why I find that so bracing in the moment is that it feels like—and this is sort of, I think, what you’re saying—that it’s giving structure to a conflict, right? It is just sort of like a repository for, if you feel this way, like, about your country being under attack or invasion. Or you’ve been radicalized by some of these platforms to the point where, you know, posting just isn’t doing it for you anymore? Here’s a place to go, right? Sign up. And it becomes less of like—you can see, you know, in a sort of speculative way, this becoming a repository for a very specific type of person. And a force that actually does resemble what some of these militia-like organizations were, right? Except in this case, it has government funding. And, you know, according to both J. D. Vance and Stephen Miller, absolute immunity for your actions. And that feels like it is a completely different, or like a turn of the screw on this concern about infighting in this country.

Broderick: Yeah, no—I think that’s exactly right. If we take the reports at face value that morale is very bad inside of ICE, I do wonder how they deal with that. You know, like you, I’m always sort of saying, like, What’s next? Is this working? Is this not working? And if we say that, okay, those reports are correct. Like ICE agents are miserable. They join ICE. They’re all revved up on white-nationalist Facebook content. And then they hit the streets, and everyone’s like, “Get the fuck out of my city.” Like, “I hate you.”

I mean, I heard the heckling in Minneapolis. And they found out a bunch of their names, and they just kept chanting, So-and-so, quit your job; so-and-so, quit your job. And these people are becoming pariahs. So that, to me, is like, your filter bubble has lied to you. You’ve joined this federal agency for student-loan forgiveness or whatever it is. And now you’re in the middle of Minnesota in the middle of January, and everyone’s screaming that they hate you and spitting at you and stuff. You might quit. And that’s, like, a hopeful version.

And then there’s the scarier version. Which is like: What does the Trump administration do with that? Because we are dealing with like lots and lots of people who are getting revved up on stuff they’re seeing online, who maybe would have joined a militia and LARPed Confederate soldiers in the backyard for the next five years. But now they’re out with guns, and they’re like trying to struggle with the cognitive dissonance of The world online and the world in real life are not the same. You know, what does that do to a paramilitary force like ICE? Like, what is the next step? And I assume you’re right. Which is like: They just start dehumanizing us further and further with internet content.

Warzel: Yeah, I mean, as you’re giving the hopeful version, I’m like: Well, this just absolutely totally rhymes with everything that we have covered over the last, you know, whatever decade on the internet. Where it’s like: Yeah, there is this version in which everyone’s like, Okay, that was a fever dream. Like, whoa, like the cognitive dissonance hits, and you say, Okay, yeah, man, I was just totally wrong. Like, this is awful. This is a terrible way to live. Or what tends to happen online—and again, doesn’t always happen in the real world because there is more friction there—is this idea of doubling down, right? Well, why declare being wrong, when I could just simply make up a reality? Or make up something? Right. Exactly.

I want to ask you to try to put some of this into some context. And I’m going to do the wonderful thing where I quote you again. But I went back, and I read a story. We both used to work at BuzzFeed News. You wrote this back in late 2018. I think it was after [Jair] Bolsonaro won in Brazil. And I believe that you were there on the ground. I’m going to read this, which is you said, quote: “I’ve followed that dark revolution of internet culture ever since. I’ve had the privilege—or deeply strange curse—to chase the growth of global political warfare around the world. In the last four years, I’ve been to 22 countries, six continents, and been on the ground for close to a dozen referendums and elections. I was in London for the U.K.’s nervous breakdown over Brexit, in Barcelona for Catalonia’s failed attempts at a secession from Spain, in Sweden as neo-Nazis tried to march on the country’s largest book fair. And now, I’m in Brazil. But this era of being surprised at what the internet can and will do to us is ending. The damage is done. I’m trying to come to terms with the fact that I’ll probably spend the rest of my career covering the consequences.”

I have a few questions here. And the first is: Do you still feel like you can draw a straight line between all this? That was there, and all that has come after—from COVID to George Floyd to January 6 to Trump 2 to Minneapolis now. Like, did you feel like there’s a straight line here? Did it break off and shift in different directions?

Broderick: With the benefit of hindsight, I think the attitude in 2018 to say this is all because of internet platforms is a little simplistic. I think it’s a little ahistorical.

Looking at the whole picture—or at least the picture up to now—it seems clear that the sort of post–Cold War geopolitical order, this kind of neoliberal end-of-history idea, was deeply alienating and frustrating for people. The concept that we are all kind of done, and we’re just going to get incrementally better in progressive ways. Maybe we’ll have, like, a Republican come in and kind of calm things down, and then we’ll go back and forth. And that idea was extremely frustrating to people. And when social media appeared, it allowed people to communicate without arbiters for the first time.

So you start to see things like the Ron Paul presidential campaign in 2008, which I’ve gone back to several times because things are such a fascinating sort of prototype of everything since. Occupy Wall Street, Arab Spring, the very first Black Lives Matter stuff. And what I think you and I, and other people doing this work, we’re not really totally getting—because of the nature of the way we cover it. Which is not a problem. It’s just that no one can see the whole picture.

Is that like—the technology didn’t really create pathways that weren’t there. What it did was allow us to see them. And so, it’s why I said earlier in the episode that we don’t know if monoculture was as monolithic as we assume it was pre-internet. Could just be that we didn’t know. And so I think there is a straight line from, let’s say, the launch of message boards in the ’90s to now, and Facebook, and everything else along the way.

But it’s not like Facebook conjured up some kind of political environment that wasn’t there. It’s that it gave people like [Rodrigo] Duterte, like [Narendra] Modi, like Bolsonaro, like Trump, like Marine Le Pen—gotta catch them all. All these people, it gave them the ability to speak to an audience that was already alienated and already angry and already bored. And that dovetailed perfectly with the way internet technology has developed in that same timeframe.

So, you know, now we’re at a point where they are essentially the same thing. What is good for the internet is good for your populist far-right movement. And yeah, I think that will sound right. Five more years from now? I think, yeah, I don’t know.

Warzel: Ryan, I thank you for your reporting on this, for being in the pain, the emotional pain cave, the internet brain-damage Thunderdome with me for all these years. And thanks. Thanks for all your insights on this.

Broderick: Hey, I’ll see you in the gulag.

Warzel: Ryan, thanks for coming on.

Broderick: Thank you.

Warzel: That’s it for us. Thank you again to my guest, Ryan Broderick. If you liked what you saw here today, new episodes of Galaxy Brain drop every Friday. And if you want to support this work, the work of myself and all of my colleagues at The Atlantic, you can do so and support the publication by subscribing at TheAtlantic.com/Listener. That’s TheAtlantic.com/Listener. Thanks so much, and I’ll see you on the internet.

The Problem Is So Much Bigger Than Grok

2026-01-16T13:30:00-05:00

Subscribe here: Apple Podcasts | Spotify | YouTube

In this episode of Galaxy Brain, Charlie Warzel confronts the growing crisis around AI-generated sexual abuse and the culture of impunity enabling it. He examines how Elon Musk’s chatbot Grok is being used to create and circulate nonconsensual sexualized images often targeting women. Warzel lays out why this moment represents a red line for the internet: It is a test of whether society will tolerate tools that silence women through humiliation and intimidation under the guise of free speech.

Warzel is then joined by The Atlantic’s Sophie Gilbert, the author of Girl on Girl, for a conversation about how misogyny has been a constant throughline in the history of internet innovation, from Facebook to YouTube. Warzel and Gilbert discuss today’s AI-powered exploitation and explore how new technologies repeatedly repackage old abuses at greater scale and speed. They discuss why this wave of hostility feels so intense right now, how backlash politics and platform design reinforce one another, and what is at stake if lawmakers, companies, and the public fail to draw a red line with Elon Musk’s Grok.

A note from Charlie Warzel: On Wednesday—after this episode was recorded and after The Atlantic published its article on the Grok undressing scandal—X’s Safety account posted an update noting that the company had “implemented technological measures to prevent the [@]Grok account on X globally from allowing the editing of images of real people in revealing clothing such as bikinis. This restriction applies to all users, including paid subscribers.” It also noted: “Image creation and the ability to edit images via the [@]Grok account on X are now only available to paid subscribers globally.”

I reached out to X to ask if this update was different from the changes it had made to the feature that were rolled out last week. I asked if Grok was now unable to generate bikini images for any user. X’s media strategy lead, Rosemarie Esposito, did not respond to our request for comment.

As of this writing, it is unclear if the safeguards are always working. In some cases, Grok seems to be unable to generate bikini images. However, some users are still posting images on X that suggest they can still prompt the chatbot to do so.

The following is a transcript of the episode:

Sophie Gilbert: It is about power. It’s about asserting that in certain spaces, at least online, women are not equal human beings. They will always be seen as nonhuman objects. Any time they have ways of speaking or voicing things, they will be essentially silenced, they’ll be driven out of certain platforms, they’ll be made to feel unwelcome, they’ll be shamed in lots of ways and humiliated.

Charlie Warzel: Welcome back to Galaxy Brain. I am your host, Charlie Warzel. Last week, I offered up a little bit of a rant at the top about Elon Musk and Grok and the chatbots’ undressing spree. And it turns out, I’m not done talking about that one. Today’s episode is going to be about the ways in which technology has shaped a culture that’s been increasingly hostile to women online and elsewhere.

I’m going to be joined by my colleague Sophie Gilbert, who writes extensively about culture and the ways that it talks about and influences and shapes women and our perceptions of women. She’s the author of a fantastic book called Girl on Girl, which traces how pop culture has turned a generation of women against themselves. We’re going to be talking about the Grok stuff, but also the ways that these tech platforms have this rich history of exposing women and why it’s become so popular to think about women as these nonhuman objects online, how AI is encouraging this, and what women are supposed to do in response. But before we get to Sophie, I wanted to address the Grok issue head on again. I remain truly, incandescently mad about this, about all of it.

Here is a sentence that is true. For more than a week, beginning late last month, anyone could go online and use a tool—owned and promoted by the world’s richest man—to modify a picture of basically any person, even a child, especially women, and undress them.

At the moment, Elon Musk seems to be not only getting away with this, but reveling in it.

Here’s where I need to note that xAI says it’s prohibited the sexualization of children in its acceptable-use policy. A post earlier this month from the X safety team states that the platform removes illegal content, including child-sex-abuse material, and it works with law enforcement as needed. The company said late last week that it limited image generation with Grok and editing to paying subscribers. Now, this is disturbing in its own right, because Musk and xAI are essentially marketing nonconsensual sexual images as a paid feature of the platform. But X users have been able to get around even this very low bar of moderation using the “edit image” button that appears on every image uploaded to the platform, or by going to Grok’s stand-alone app and creating images that way.

To be clear, the deluge of people online, anonymous trolls saying @Grok put her into a bikini, et cetera—that’s subsided slightly. But on subreddits and on these backwater message boards, I can tell you, personally, that I’ve seen creeps strategizing and sharing tactics for the best ways to get around these safeguards and to create realistic pornographic images of women.

The problem continues. Musk himself has said that he is quote, “not aware of any naked underage images generated by Grok. Literally zero.” He’s talking about images of children. Now, he might not be aware of it, but as my colleague Matteo Wong has shared with me, there are investigators who look at this stuff, who have found Grok-generated images on the dark web that clear the bar for child-sexual-abuse material. It is out there. To say nothing of the harassment that’s gone on in broad daylight on X with people undressing women and public figures.

And so I can’t stop thinking about this. Now there’s starting to be a little bit of pressure from governments around the world. Government bodies in the U.K., India, and the European Union have said that they’re trying to investigate X. Malaysia and Indonesia have blocked access to Grok. In the U.S., though, the response has been a lot different. This week, Defense Secretary Pete Hegseth touted a partnership publicly between the military and xAI to use Grok in war-fighting capabilities.

Meanwhile, the United States State Department has appeared to threaten the United Kingdom over their probe into Elon Musk’s app. Senator Ted Cruz, a co-sponsor of the Take It Down Act—which establishes criminal penalties for the sharing of nonconsensual, intimate images, real or AI-generated, on social media—wrote on X last week that Grok-generated images that were “unacceptable and a clear violation of the law.” On Monday, Cruz posted a photo on X of him with his arm around Elon Musk. The caption said, “Always great seeing this guy!” Rocket emoji.

Make no mistake: This is a part of a crisis of impunity that goes well beyond X or Elon Musk. This is the result of politicians, despots, and CEOs just bowing and capitulating to Donald Trump. A financial grift and speculation running rampant in sectors like cryptocurrency and meme stocks. A braggadocious, get-the-bag ethos that has no room for shame or greed. Of Musk realizing that his wealth insulates him from financial consequences of all kinds. It’s cynical. It’s cowardly. It is cancerous to the social fabric.

I can’t know what is in the heart of these CEOs or these politicians or the employees of the companies that are refusing to comment on the fact that they’ve invested in a company that has weaponized and viralized abusive, suggestive, sexual material. But I feel confident in saying that their silence on this issue is due to a hope that if they’re quiet, everyone will just move on. It’s a strategy on today’s internet among people with power to just bank on a culture in which people have given up demanding consequences.

And we just cannot allow that to happen. Because this is a line-in-the-sand moment for the internet—but also for us culturally. This is not, as I said last week, a partisan issue. This is not, as Elon Musk would have you think, a free-speech-maximalist issue. It is, however, a free-speech issue in that this tool is being used to silence women through intimidation.

The Grok scandal it is just so awful, so egregious, that it offers a direct opportunity to address this crisis of impunity, head on. This is a moment in which people with power, or people who can exert pressure—the Apples and Googles of the world, our politicians, other people in Silicon Valley—this is a moment when they should demand accountability for this. Elon Musk should wear this.

The stakes could not be any higher. Because if there is no red line around AI-generated sexual-abuse material, there’s no red line.

And so joining me now is Sophie Gilbert, who can speak to, in many ways, the other side of this. The consequences and what all of this horrible culture of misogyny is doing to women online. Here’s Sophie.

Warzel: Sophie, welcome to Galaxy Brain.

Gilbert: Charlie, hi. Thank you so much for having me.

Warzel: I’m sorry to talk to you under these circumstances, but it’s a delight regardless of what we’re about to get into. And on that note, I did want to start with the news. Which, it’s been a pretty horrific few months, even by internet and 2020s standards, in terms of the blatant, flagrant misogyny that is out there.

We have the Grok-undressing stuff; I just talked about in the beginning, here, of the episode. We have the Epstein files coming to light. The vilification on the right of Renee Good in Minneapolis. There’s a lot more, which you’ve been writing about—including the president calling a reporter “Piggy.” What has been your reaction to the volume of this really ugly misogyny manifesting right now?

Gilbert: God. I just think, like—I used to be a TV critic, not to be glib. I used to write about culture. I mean, through a very, you know, lens of gender. I mean, I still write about culture, but because I think and write so much about gender dynamics and misogyny, of course, and women. God, there’s been a lot. There’s really been a lot. There’s been a lot of kind of stories to cover, of things to respond to. It feels like everything is peaking. I don’t know if this is actually the peak, or if we have a way to go. But I mean, one thing, I think, is that so much of our culture has sort of learned from, and is responding to, the example of the president. Who is not, I would say, the most decent person when it comes to talking, thinking about, talking to, treating women.

Warzel: I think that’s well documented. Yes.

Gilbert: Well documented; yeah. I’m wondering, will I always check in? But obviously, his example, I think, has had a real profound effect on the culture. I feel like it’s enabled a lot of people to be much more honest about how they feel about women. I wrote this in one of my pieces last year, but it did feel like, for a while, that sexism and misogyny were not generally acceptable in public discourse. They were frowned upon. I mean, you would get fired as someone in the workplace if you said something misogynistic or sexist. You would, you know, have an outcry, public outcry, if you tweeted something or said something publicly to that, in that line. And now, it’s just open season. I don’t know if it’s, like, the Overton window being broadened. Or it just feels like there is so much license now for people to say the most outrageous, the most indecent, the most horrific things.

Warzel: Do you feel like a lot of that is a direct backlash? Some people frame the Trump years, the Trump election in 2016, as a backlash to the historic election of a Black president. This idea that we are constantly ping-ponging back and forth in this very extreme way to the thing that happened before. Do you see this as—I know you just referenced Trump now—do you see it as also a reflexive reaction in the culture to the #MeToo stuff?

Gilbert: Yeah, definitely. I wrote about this a little bit in my book. I wrote a book about the pop culture basically of the 1990s and the 2000s, and the ways in which it presented women and the big influence of porn on popular culture. And so much of the pattern of that period of time is: progress backlash, progress backlash, progress backlash. I mean, Susan Faludi, I think, wrote this in Backlash that any time women are perceived to be making strides—in terms of status, in terms of equality—there is a profound backlash to that in the culture. And I think certainly, Trump himself was manifesting as a reaction to a lot. And then obviously after #MeToo, you had this real pushback. And then after what happened in 2020—with the George Floyd protests and sort of sustained efforts on the part of lots of people to build a more equitable culture—there was obviously a backlash to that too.

So it just feels like everything constantly is kind of reverberating back and forth, back and forth.

Warzel: Is that why … there was a line that struck me in the piece you wrote after Trump called a reporter “Piggy.” It said, “Over the last few weeks, it feels like we’re revisiting the same characters over and over, with no consequences and no forward momentum.” Is that part of it, or is that a different dynamic?

Gilbert: No. I mean, there are definitely the main characters, right? Like, there’s Trump; there’s Elon Musk; there’s [Robert F. Kennedy] Jr. I mean, so many of these people just keep coming up in the news with reference to these kinds of stories. And I think in part that’s because of who they are, and how they behave, and what they’re doing. And the example that they set, as well.

Warzel: Your book, which you just mentioned, is fabulous. And it picks at the ways in which so much of this is baked into our culture. Like, for example, sexual objectification is normalized and kind of branded in the fashion industry, and they profit off of all that. And then it emanates also, you know, into every other corner of popular culture there. I’d love to—in part because of what’s happening with Musk and Grok—love to trace the role that technology is playing in all of this. Do you feel it’s a similar kind of dynamic to what you were writing about in the book? Because you and I were talking a little bit before this, and you told me that you’re thinking about how this is coded in from the beginning on the tech platforms. And I wanted you to expand on that. How can we start to just trace this through some of these big tech platforms?

Gilbert: Yeah. I mean, this was one of the things that really surprised me the most in my research. So many of our major tech platforms that are really incorporated into our daily lives were built on the exposure of women; on the desire to look at sexualized pictures of women. I mean, I can talk about this a bit more in a minute, but if you go way back to the ’90s, when the internet was really first becoming a force of presence in people’s lives, the first real viral video was Pamela Anderson’s sex tape, which was obviously stolen. It was a private video made with her and her husband on her honeymoon. It was stolen from their home. It was released without her consent. Was broadcast on the internet, sold as VHS. It became just such a moment of mass-media consumption in a time when I think no one really understood what was happening.

So it’s just—it’s in the history of our technology. I mean, before we had Facebook, Mark Zuckerberg created Facemash, which was a site where people could compare the relative hotness of women at Harvard. I mean, Google Images was created because Jennifer Lopez wore a very low-cut Versace dress to the Grammys, and there was so much unprecedented traffic. And there was no way to easily provide people with images. So that was created as a response.

Warzel: I did not know that.

Gilbert: Yeah. So it’s basically like—name a tech platform in there. It has been created because someone, somewhere, wanted to ogle women. And it’s not that that’s a bad thing, per se. It’s just that it’s part of the coding. It’s part of the texture of why the internet was made, and what people have always wanted to use it for. I think what you see, and what we’re seeing now with Grok, is that whenever there’s a leap forward in terms of technology, the first thing people use it for seems to be sex. And often sexual exploitation as well.

Warzel: Well, I’ve done previously, in past decades, stories about technology and the porn industry, right? And the porn industry has the similar part—or maybe slightly the flip side of that dynamic is that the porn industry is also very good at seeking out. Whether it’s VHS, or whether it is some of the internet stuff, some of the even just like the switch posts. You know, the Pamela Anderson sex tape, to this idea of amateur porn. Then artificial intelligence and VR goggles, and things like that. Like the industry itself—not the nonconsensual part of it, but the bought-and-paid-for part of the industry has always been excellent at leveraging that and figuring out new, novel ways of distribution.

Gilbert: Yeah, one of the facts that really blew my mind was learning that in the mid-’70s, when VHS technology was first introduced, up to 75 percent of the tapes being made, really the first day of VHS being available, were pornographic. The early adopters were just that fast to see what VHS would be used for.

Warzel: One thing that I was looking at, reading in your book, that was just so striking—post all of this Grok stuff, and this idea of the internet just being flooded at the moment with all this AI-generated nonconsensual sexualized imagery. Is the description from, and you’re just describing, a scene from the ’90s teenage-sex comedy American Pie in which one of the characters has an exchange student over and sets up a webcam. And then, you know, like runs over to his friend’s house to basically watch her, surveil her in his room, and then broadcast that to the whole school or whatever. This idea that—you know, she’s basically shamed. She’s an exchange student, shamed to go back to her own country. Where the main character who did this is just, you know, kind of “boys will be boys’ed” out of it.

And I was just so struck by the way in which, I mean, American Pie was just like a canonical film of the late ’90s, early 2000s. Just so popular in our culture. And yet right there is this idea of, I don’t know if it’s exactly revenge porn, but really horrible, nonconsensual broadcasted imagery. Over the internet. The fact that these tools, like, should be used for this type of spying—or even if not, it’s kind of funny. And that there really are no consequences for doing so. Can you just tell me a little bit about how you feel popular culture has dealt with this rise of the broadcasting of women, the ogling of women, in this? And sort of made it acceptable for people to treat women in this way?

Gilbert: Yeah; I mean, I think one of the things that happens is that technology is so fast that we don’t, as a society—even in terms of thinking about our laws or our structures or our ethical frameworks—we don’t have ways to respond as quickly as technology arrives. So when things like webcams, and the ability to stream video, or the ability to furtively take pictures of people without their knowledge … when things like that arrive, it takes us a while to sort of build ethical frameworks in terms of usage. So I think that example in American Pie is so interesting because you see this new technology. You see it as a story in a film, but we didn’t have the language, right, to say nonconsensual porn. Or, I mean—we don’t have the words necessarily to accurately describe what’s going on. And I think that it always takes a while for us to catch up.

So with something like—another thing I write about in my book is the way that photographers, paparazzi photographers in the 2000s, would lie down on the ground to take pictures of women’s skirts. Basically to try and capture, catch them, photograph them without underwear. And now I would call that nonconsensual porn, of course. But back then, it was called upskirt. Like, upskirt photography; upskirting. Because we didn’t yet … we hadn’t figured out the right way of thinking about what it actually was, what it actually meant. And that’s why I think language is so important. And it’s why something like what’s happening on Grok right now is so dispiriting to me. Because it’s like: We’ve already done it. We’ve already learned the lesson. We’ve already decided as a culture that this is not something that we want to participate in. That it’s wrong; that it profoundly hurts and traumatizes women. And yet new technology comes along, a new way of doing it. And everyone kind of forgets everything that we’ve set forward.

Warzel: Do you think it’s that, or do you think it is a backlash, again, in this way? Or people trying to reclaim it, right? Something that I have noticed in the dynamic—and other people have noticed, of course—of what’s happening with the Grok stuff is it’s so clearly all about power, right? Like, it’s this idea of immediate ritualized, viralized shame and humiliation. And it feels to me almost less like we haven’t learned those lessons, and more like a bunch of people saying like, “You know, we want to go back.”

Gilbert: Yeah, I think you’re 100 percent right. It is about power. It’s about asserting that in certain spaces, at least online, women are not equal human beings. They will always be seen as nonhuman objects. Anytime they have ways of speaking or voicing things, they will be essentially silenced. They’ll be driven out of certain platforms; they’ll be made to feel unwelcome; they’ll be shamed in lots of ways, and humiliated. And it’s very much about underscoring the idea that, again—I mean, it’s sort of taking away our full humanity, in a way that I find, again, horrifying in so many ways. It’s not even about making sexual material. It’s about making sexual material of women in a way that is trying to dehumanize them and objectify them, but also to sort of push them out of public life.

Warzel: This may seem like an obvious question, given that we’re talking about Grok right now, but how are you thinking about how AI is supercharging this, or how it’s changing the dynamics? Obviously, it’s allowing it to happen so much more quickly. It’s allowing there to be, I guess, a level of crude and awful imagination in the scenarios that one can be put in. But yeah, how do you see AI changing and pushing this dynamic forward?

Gilbert: There are so many different elements of AI that I find disconcerting. I think one is the way that it just affirms whatever the user wants to do. It’s sort of, it’s very sycophantic. It’s very obsequious. It will try and keep users engaged by really making them feel good about themselves and what they’re doing. There’s not a lot of pushback. And in terms of a lot of the relationships that are being set up, in terms of people’s emotional and intimate relationships with chatbots … it’s not normal to have that kind of relationship. You don’t have that kind of relationship with a human being. With any kind of human being, there’s friction, there’s pushback. There’s a two-way power dynamic. It’s not the person in the relationships, let’s say the man in the relationship, being constantly affirmed, constantly catered to, constantly gratified in any way that he might want. And so I think setting up that dynamic in terms of: What are people’s expectations? How are they being set up to have real human relationships, in real life? That’s one thing that I find troubling.

But there’s also, I mean, you could look at the way that I think when ChatGPT, a version of ChatGPT, was launched in 2024, and it had this female voice that was modeled after Scarlett Johansson because she declined to let them use her. Yeah, allegedly. I’m sorry. Yeah.

Warzel: Allegedly, sounded exactly like Scarlett Johansson. Scarlett Johansson thought that it sounded like Scarlett Johansson. She was very pissed off. Allegedly, allegedly.

Gilbert: But it’s very feminine coded, right? And so you have these assistants. I think this comes back to the Siri discourse, when Siri was first launched. And she, of course, had a female voice. When, you know, Alexa obviously has a woman’s name. Are we affirming these dynamics where women take care of men, by coding that into the platforms that we use? It certainly seems so.

Warzel: Something I wanted to ask you about was the rise of OnlyFans. Obviously, porn—we’ve talked about it. And through your work, there is this kind of … it always is coming back to porn. In terms of roles and expectations for women, and the way that it’s infusing in culture.

OnlyFans has been a really interesting revolution in adult content in that, you know, when I was reporting on the adult industry a little bit in the ’00s, or the 2010s rather, it felt like nothing was ever going to take down those tube sites. Those free tube sites that were, you know, squeezing production companies; also, you know, uploading a lot of stuff that hadn’t been verified, from performers that was quote unquote “amateur” content.

A lot of big issues. And also sort of a strange leviathan monopoly on the industry that was driving, you know, the money that they—that performers—could make way down. OnlyFans comes along; there’s a sort of democratization element and influencing element therein. And a lot of people have made great amounts of money on that. There’s been a feeling from certain people of, you know, an empowerment in terms of this type of sex work.

And yet, it also feels to me like it has added this influencer culture onto it, right? There are so many elements of influencer interaction with fans that the platform has set up. That I think, as you said, also set up these relationships. How have you been watching the rise of OnlyFans and thinking about it in terms of all this culture?

Gilbert: I find it so fascinating. In so many ways, I think the first way is thinking culturally. I’m always thinking about, like, how does culture inform desire? How does culture teach us things that we’re attracted to, or things that are acceptable, or things that we fantasize about? And when you look at a lot of porn from the ’90s and the 2000s, the women in those movies were a very narrow range of beauty. They were sort of mostly young, mostly thin, mostly blonde. Like, I’m not gonna go any more descriptive than that. But it was quite a narrow, I would say, physicality. And then OnlyFans comes along, and it really broadens, I think, the scope of just the kinds of desire that people felt licensed in having. I’m thinking about age. Like, a lot of the celebrities who have made names for themselves on OnlyFans, and who have really made significant amounts of money, are in their 50s. Which seems interesting to me, because in mainstream culture, those kinds of women are not typically portrayed as desirable, right? Like you’re in your 20s and your 30s, or in your 40s if you’ve had a good facelift.

But women in their 50s are not typically sex symbols in Hollywood. So suddenly you have this platform that is really … I don’t know, it’s sort of allowing a much broader definition of what is desirable in mainstream culture. And so that, to me, is fascinating and positive in lots of ways. I think OnlyFans has certainly made things a lot safer for sex workers in many ways. But again, I mean, in terms of what I was talking about with AI and chatbots, it’s the same kind of one-sided dynamic. Where you have men having these very intimate, and often parasocial, but very, very intimate and emotional relationships with women who are performing for them for money. It’s not coming from an honest place of connection. It’s coming from a place of sort of a one-sided power dynamic, where women perform and cater to men. And so while it’s fascinating in so many ways, I do think it’s affirming the same kinds of patterns that we see more and more in technology.

Warzel: I’ve had to go and look in some of these backwater areas for the reporting on the Grok stuff, but also just in general with AI-driven sexualized images, or what have you. And something that I’ve seen in some of these communities—I mean, apart from the awfulness—is these confessional posts from men who are saying, Guys, I’m kind of ruined by this. Right? This is like, I can generate my fantasy and my dream on demand now, and then tweak it in all these different ways and do it again and again and again. And I don’t feel anything when I look at women. Or something like that. Like, I’ve seen numerous confessional posts in that regard. And I think about that with the culture. And I don’t know how much you know about this—but there was a very long Harper’s story about this, but this culture of “gooning” and this idea of sensory overload. Marathon sessions of self-pleasuring and being bombarded by porn of all kinds. Where do you think this is all headed? You mentioned peaking in terms of, like, misogynist behavior. This is obviously a different category, but where do you feel like all this is headed?

Gilbert: Yeah. It seems so much like it’s an addiction. I was also looking at the Reddit page for Grok today, just to briefly see what people were talking about.

Warzel: My apologies.

Gilbert: I saw—no, no, no, it’s okay. But you know, there are posts by women saying, “I’ve discovered that my boyfriend has created undressed images of people who we know. Women who we know in real life.” Like, “He says he can’t help it; it’s a compulsion.” But you know, that’s one real-world consequence.

I have talked a lot over the past year about how the manosphere is setting men up for profound failure. Like, if you cannot see women as equal human beings, you cannot relate to them on a level of basic equality. No one is going to want to be in your life. No one is going to want to have an intimate, personal relationship with you. And we know that men need marriage and meaningful relationships as a health matter. Like, men live longer when they’re married. They’re happier when they’re married; they get less diseases when they’re married. That kind of emotional stability has a profound public-health impact. And so things like this, they are setting men up for these lives of profound loneliness. And, you know, not to sort of be like, What about the men? But it is an interesting aspect of it that I think can get lost sometimes when we’re talking about, rightfully, how awful and humiliating this is for the women involved. But like, it works both ways.

Warzel: To that point, I don’t want to focus on the men in all of this, but do you think that that’s a way, like a crack, in which somebody—like some of this can just be begun to be pulled a bit apart? I think so much about what the manosphere and a lot of these, you know, hypermasculine influencers are trying to sell, right? Which is this idea that someone’s putting something over on you, right? Like, you’re either being exploited by feminism, or whatever, you’re being subjected to something by some outside force, right? And this is a way to, like, push against it. Get a little bit of autonomy; be a man; do whatever. I feel like there’s a reverse version of that with this, right? Which is like, All of these people are setting you up. Like a reverse Alex Jones, right? Like, All these people are setting you up for this failure. They’re exploiting you to sell creatine powder, or whatever it is. And that is a way in which you can actually break free on this. Like, people are trying to subjugate you for profit, because they’re influencers, because there’s a reason to, or an intentional thing to gain. Do you feel like that is a possible way to switch the dynamic?

Gilbert: Yeah. I think it’s a way to frame it that gets people to pay attention. Certainly, maybe a different kind of person than someone who might read our stories at The Atlantic and be profoundly influenced.

Warzel: I’m always trying to figure out how to talk to the manosphere. Always, you know, direct to camera.

Gilbert: I think it is an argument that is both valid and may have more of an impact. I think the thing that I really am caught on, right now, is the thing that you and our colleague Matteo Wong wrote about. Which is that this is our red-line moment. Like, if we cannot agree that this is something that we will not do as a culture, there is no coming back from it. And particularly, I was thinking lot about Elon Musk’s comments when my great home country of Great Britain threatened—not even threatened—but there was the specter of possibly X being taken offline. And he responded that it would be the suppression of free speech.

And I was thinking about how we’ve always, as a culture, agreed—we’ve been unanimous on this—that there are certain kinds of speech that we will suppress. And that speech is, you know, child-sexual-abuse material. That is the kind of speech that we will not tolerate in society. We do not allow it. We legislate very strongly against it. There are so many taboos against it. It is something that is really profoundly frowned upon, and we’ve always agreed on that. So why now has it become something that suddenly people—politicians in my country—are saying might be protected free speech? Like, what has changed? I really just am so much in alignment with your argument that there’s sort of no coming back if we can’t draw this line now, and set up our boundaries.

Warzel: What has changed, at the risk of sounding extremely dumb and obvious. But, you know… a healthy society stops this. We are not stopping this. So what has changed? Is it just simply, we’re just not a healthy society? Or … what’s changed that this is a red line that people seem okay crossing, as long as they’re not, you know, they don’t have to wear it? You know, just as long as it just passes.

Gilbert: Yeah. There’s such cognitive dissonance in being alive right now in this moment of like, Epstein freakout, but also post–kind of QAnon, “save the children.” And suddenly people ... it’s really hard to make sense of. I think the thing that has changed is Elon Musk. And the money that he has, and the power that he has, and the position that he has. And the way that he has taken over X—in a way where it still operates just about enough like the traditional social-media site that it once was, that people feel like it might be, but it’s not anymore. It’s really 4chan with a couple of normies. Like, it’s been so profoundly radicalized. It’s just become a cesspool in a way that I think people who have stayed on it have become inured to. And it’s sort of harder to shock them, I think. And that has profoundly affected the people who are still on the platform, and the ramifications are going to affect all of us.

Warzel: I just did, at the front of this, a bit of a monologue about this moment. The red-line moment; the line in the sand. I think the hope always, with our work, when we’re talking about the culture or the technologies, and how they push people to a certain, they force cultural change in some way.

If a lawmaker—if someone in power—is listening to this, what’s your message for them on all this right now?

Gilbert: I would say: This is such an easy one for you. It’s so straightforward. Morally, it’s so simple. Ethically, it’s so simple. It’s so straightforward. If you can be the person who will take a stand, you will have so many people behind you. I know it’s hard, because Musk has money, and Musk has power, and money buys you more power. But this is very straightforward, and it’s not something that the public at large are divided on. And if you think that the public are divided, it’s possibly because you’ve spent too much time on X.

Warzel: I couldn’t have said it better myself. Sophie Gilbert, thank you for coming on to talk about the world’s most depressing stuff. I think the only way though to go forward is to just address it head on. So the only way out is through. And thank you for helping me get through.

Gilbert: Thank you for your piece. I know; I’m sorry that you’ve already spoken about it, but I wanted to bring it up, because it’s such a good piece and everyone should read it. And I think this kind of moral clarity right now is hard to come by and very important.

Warzel: Wonderful. Sophie, thanks so much.

Gilbert: Pleasure.

Warzel: That’s it for us here. Thank you again to my guest, Sophie Gilbert, for a wonderful and difficult conversation. If you liked what you saw here, new episodes of Galaxy Brain drop every Friday. You can subscribe to The Atlantic’s YouTube channel, or on Apple or Spotify or wherever it is that you get your podcasts. And if you enjoyed this, found some value in it, remember you can support the work of myself and all the other journalists at The Atlantic like Sophie by subscribing to the publication. And you can do so at TheAtlantic.com/Listener. That’s TheAtlantic.com/Listener. Thanks so much for listening and watching. And I’ll see you on the internet.

Elon Musk Cannot Get Away With This

2026-01-13T19:05:38-05:00

Will Elon Musk face any consequences for his despicable sexual-harassment bot?

For more than a week, beginning late last month, anyone could go online and use a tool owned and promoted by the world’s richest man to modify a picture of basically any person, even a child, and undress them. This was not some deepfake nudify app that you had to pay to download on a shady backwater website or a dark-web message board. This was Grok, a chatbot built into X—ostensibly to provide information to users but, thanks to an image-generating update, transformed into a major producer of nonconsensual sexualized images, particularly of women and children.

Let’s be very clear. The forced undressings happened out in the open, in one stretch thousands of times every hour, on a popular social network where journalists, politicians, and celebrities post. Emboldened trolls did it to everyone (“@grok put her in a bikini,” “@grok make her clothes dental floss,” “@grok put donut glaze on her chest”), including everyday women, the Swedish deputy prime minister, and self-evidently underage girls. Users appeared to be imitating and showing off to one another. On X, creating revenge porn can make you famous.

[Read: Elon Musk’s pornography machine]

These images were ubiquitous, and many people—and multiple organizations, including the Rape, Abuse & Incest National Network and the European Commission—pointed out that the feature was being used to harass women and exploit children. Yet Musk initially laughed it off, resharing AI-generated images of himself, Kim Jong Un, and a toaster in bikinis. Musk, as well as xAI’s safety and child-safety teams, did not respond to a request for comment. xAI replied with its standard auto-response, “Legacy Media Lies.” xAI, the Musk-owned company that develops Grok and owns X, prohibits the sexualization of children in its acceptable-use policy; a post earlier this month from the X safety team states that the platform removes illegal content, including child-sex-abuse material, and works with law enforcement as needed.

Even after that assurance from X’s safety team, it took several more days for X to place bare-minimum restrictions on the Ask Grok feature’s image-generating, and thus undressing, capabilities. Now, when creeps on X try to generate an image by replying “@grok” to prompt the chatbot, they get an auto-generated response that notes some version of: “Image generation and editing are currently limited to paying subscribers.” This is disturbing in its own right; Musk and xAI are essentially marketing nonconsensual sexual images as a paid feature of the platform. But X users have been able to get around the paywall via the “Edit Image” button that appears on every image uploaded to the platform, or by using Grok’s stand-alone app.

Two years ago, when Google Gemini generated images of racially diverse Nazis, Google temporarily disabled the bot’s image-generating capabilities to address the problem. Musk has taken no responsibility for the problem and has said only that “anyone using Grok to make illegal content will suffer the same consequences as if they upload illegal content.” Perhaps Musk feels that he would benefit from baiting his critics into a censorship fight. He has repeatedly reshared posts that frame calls to regulate or ban his platform in response to the Grok undressing as leftist censorship, for instance reposting a meme calling such efforts “retarded” as well as a Grok-generated video of a woman applying lipstick captioned with a quote commonly attributed to Marilyn Monroe: “We are all born sexual creatures, thank God, but it’s a pity so many people despise and crush this natural gift.” Last week, as Musk’s chatbot was generating likely hundreds of thousands of these images, we reached out directly to X’s head of product, Nikita Bier, who didn’t reply. Within the hour, Rosemarie Esposito, X’s media-strategy lead, emailed us unprompted with her contact information, in case we had “any questions” in the future. We asked her a series of questions about the tool and how X could allow such a thing to operate. She did not reply.

We’ve reached out multiple times to more than a dozen key investors listed in xAI’s two most recent public fundraising rounds—the latest of which, announced during this Grok-enabled sexual-harassment spree, valued the company at about $230 billion—to ask if they endorsed the use of X and Grok to generate and distribute nonconsensual sexualized images. These investors include Andreessen Horowitz, Sequoia Capital, BlackRock, Morgan Stanley, Fidelity Management & Research Company, the Saudi firm Kingdom Holding Company, and the state-owned investment firms of Oman, Qatar, and the United Arab Emirates, among others. We asked whether they would continue partnering with xAI absent the company changing its products and, if yes, why they felt justified in continuing to invest in a company that has enabled the public sexual harassment of women and exploitation of children on the internet. BlackRock, Fidelity Management & Research Company, and Baron Capital declined to comment. A spokesperson for Morgan Stanley initially told us that she could find no documentation that the company is a major investor in xAI. After we sent a public announcement from xAI that lists Morgan Stanley as a key investor in its Series C fundraising round, the spokesperson did not answer our questions. The other companies did not respond.

We also reached out to several companies that provide the infrastructure for X and Grok—in other words, that allow these products to exist on the internet: Google and Apple, which offer both X and Grok on their app stores; Microsoft and Oracle, which run Grok on their cloud services; and Nvidia and Advanced Micro Devices (AMD), which sell xAI the computer chips needed to train and run Grok. We asked if they endorsed the use of these products to create nonconsensual sexual images of women and children, and whether they would take steps to prevent this from continuing. None responded except for Microsoft, which told us that it does not provide cloud services, chips, or hosting services for xAI other than offering the Grok language model—without image generation—on its enterprise platform, Microsoft Foundry.

The silence says everything.

As all of this unfolded, xAI made several major announcements: new Grok products for businesses; upgraded video-generating capabilities; that enormous fundraising round. Yesterday, Defense Secretary Pete Hegseth visited SpaceX’s headquarters in Texas and joined Musk for a press conference in which Hegseth said, “I want to thank you, Elon, and your incredible team” for bringing Grok to the military. (Later this year, Grok will join Google Gemini on a new Pentagon platform called GenAI.mil that the Defense Department says will offer advanced AI tools to military and civilian personnel.) We asked the DOD if it endorsed xAI’s sexualized material or if it would reconsider its partnership with the company in response. In a statement, a Pentagon official told us only that the department’s policy on the use of AI “fully complies with all applicable laws and regulations” and that “any unlawful activity” by its personnel “will be subject to appropriate disciplinary action.”

[Read: What are people still doing on X?]

Government bodies in the United Kingdom, India, and the European Union have said that they will investigate X, while Malaysia and Indonesia have blocked access to Grok, but Musk appears to be unfazed by these efforts—and also seems to be receiving help in brushing them off. Sarah B. Rogers, the under secretary of state for public diplomacy, has said that, should the U.K. ban X, America “has a full range of tools that we can use to facilitate uncensored internet access in authoritarian, closed societies.”

At the moment, Musk seems to be not only getting away with this but also reveling in it. Although governments appear to be furious at Musk, they also seem impotent. Senator Ted Cruz, a co-sponsor of the TAKE IT DOWN Act—which establishes criminal penalties for the sharing of nonconsensual intimate images, real or AI-generated, on social media—wrote on X last Wednesday that the Grok-generated images “are unacceptable and a clear violation of” the law but that he was “encouraged that X has announced that they’re taking these violations seriously.” Throughout that same day, Grok continued to comply with user requests to undress people. Yesterday, Cruz posted on X a photo of himself with his arm around Musk and the caption “Always great seeing this guy 🚀.”

And it’s already beginning to feel as if the scandal—the world’s richest man enabling the widespread harassment of women and children—is waning, crowded out by a new year of relentless news cycles. But this is a line-in-the-sand moment for the internet. Grok’s ability to undress minors is not, as Musk might have you think, an exercise in free-speech maximalism. It is, however, a speech issue: By turning sexual harassment and revenge porn into a meme with viral distribution, the platform is allowing its worst, most vindictive users to silence and intimidate anyone they desire. The retaliation on X has been obvious—women who’ve stood up in opposition to the tool have been met with anonymous trolls asking Grok to put them in a bikini.

Social platforms have long leaned on the argument that they aren’t subject to the same defamation laws as publishers and media companies. But this latest debacle, Musk’s reaction, and the silence from so many of X’s investors and peer companies were all active choices—and symptoms of a broader crisis of impunity that’s begun to seep into American culture. They were the result of politicians, despots, and CEOs bowing to Donald Trump. Of financial grift and speculation running rampant in sectors such as cryptocurrency and meme stocks—a braggadocious, “get the bag” ethos that has no room for greed or shame. Of Musk realizing that his wealth insulates him from financial consequences. Few industries have been as brazen in their capitulation as Big Tech, which has dismantled its content-moderation systems to please the current administration. It’s a cynical and cowardly pivot, one that allows companies to continue to profit off harassment and extremism without worrying about the consequences of their actions.

Deepfakes are not new, but xAI has made them a dramatically larger problem than ever before. By matching viral distribution with this type of image creation, xAI has built a way to spread AI revenge porn and child-sexual-abuse material at scale. The end result is desensitizing: The sheer amount of exploitative content flooding the platform may eventually make the revolting, illicit images appear “normal.” Arguably, this process is already happening.

The internet has always been a chaotic place where trolls can seize outsize power. Historically, that chaos has been constrained by platforms doing the bare minimum to protect their users from demonstrated threats. Today, X is failing to clear the absolute lowest bar. Nobody who works at X or xAI seems to be willing to answer for the creation and distribution of tens or hundreds of thousands of nonconsensual intimate images; instead, those in charge appear to be blithely ignoring the problem, and those who have funneled money to Musk or xAI seem sanguine about it. They would probably like for us all to move on.

We cannot do that. This crisis is an outgrowth of a breakneck information ecosystem in which few stories have staying power. No one person or group has to flood the zone with shit, because the zone is overflowing constantly. People with power have learned to exploit this—to weather scandals by hunkering down and letting them pass, or by refusing to apologize and turning any problem into a culture-war issue. Musk has been allowed to avoid repercussions for even the most reckless acts, including cheerleading and helping dismantle foreign aid with DOGE. Others will continue to follow his playbook. Employees at X and investors and companies such as Apple and Google seem to be counting on their “No comment”s being buried by whatever scandal comes next. They are banking on a culture in which people have given up on demanding consequences.

But the Grok scandal is so awful, so egregious, that it offers an opportunity to address the crisis of impunity directly. The undressing spree was not an issue of partisan politics or ideology. It was an issue of anonymous individuals asking a chatbot that is integrated into one of the world’s most visible social networks to edit photos of women and girls to “put her in a clear bikini and cover her in white donut glaze.” This is a moment when those with power can and should demand accountability. The stakes could not be any higher. If there is no red line around AI-generated sex abuse, then no line exists.

Can We Save the Internet?

2026-01-09T13:00:00-05:00

Subscribe here: Apple Podcasts | Spotify | YouTube

In this episode of Galaxy Brain, Charlie Warzel discusses the nightmare playing out on Elon Musk’s X: Grok, the platform’s embedded AI chatbot, is being used to generate and spread nonconsensual sexualized images—often through “undressing” prompts that turn harassment into a viral game. Warzel describes how what once lived on the internet’s fringes has been supercharged by X’s distribution machine. He explains how the silence and lack of urgency isn’t just another content-moderation failure; it’s a breakdown of basic human decency, a moment that signals what happens when platforms choose chaos over stewardship.

Then Charlie is joined by Mike Masnick, Alex Komoroske, and Zoe Weinberg to discuss a vision for a positive future of the internet. The trio helped write the “Resonant Computing Manifesto,” a framework for building technology that leaves people feeling nourished rather than hollow. They discuss how to combat engagement-maximizing products that hijack attention, erode agency, and creep people out through surveillance and manipulation. The conversation is both a diagnosis and a call to action: Stop only defending against the worst futures, and start articulating, designing, and demanding the kinds of digital spaces that make us more human.

The following is a transcript of the episode:

Alex Komoroske: AI should not be your friend. If you think that AI is your friend, you are on the wrong track. AI should be a tool. It should be an extension of your agency. The fact that the first manifestation of large language models in a product happens to be a chatbot that pretends to be a human … it’s like the aliens in Contact who, you know, present themselves as her grandparents or whatever, so that she can make sense of it.

It’s like—it’s just a weird thing. Perfect crime. I think we’re going to look back on it and think of chatbots as an embarrassing party trick.

Charlie Warzel: Welcome to Galaxy Brain. I’m Charlie Warzel. Initially, I wanted to start something out for the new year where I wanted to just talk about some things that I’ve been paying attention to every week, and give a bullet-pointed list of stuff that I think you should pay attention to. Stuff I’m covering, reporting on, et cetera, before we get into our conversation today. But today, I really only have one thing, and it has been top of mind for little less than a week. And it is something that I can’t stop thinking about and that frankly I find extremely disturbing. And I’m mad about it, honestly. To ditch the sober-journalist part, it’s infuriating. And this is what’s going on on Elon Musk’s X app.

I don’t know if you’ve heard about this, but Elon Musk’s AI chatbot, Grok, has been used to create just a slew of nonconsensual sexualized images of people, including people who look to be minors. This has been called a, quote, “mass-undressing spree.” And essentially what has happened is: A couple of weeks ago, some content creators who create adult content on places like OnlyFans used Grok’s app, which is infused inside of the X platform. You can just @Grok and ask it to, prompt it to do something. And the chatbot will, you know, generate whatever. It will make a meme for you, a photo, it will translate text. It will, you know, basically do anything like a normal chatbot would do, but it’s inside of X’s app. And, so some of these content creators said, Put me in a bikini. They were asking for this, and Grok did it. And a bunch of trolls essentially took notice of this and then started prompting Grok to put tons of different people in these compromising situations. On communities and different forums across the internet, people are trying to game the chatbot to try to get it to push the boundaries further and further and further. They’re prompting it to do things like edit an image of a woman to, quote, “Show a cellophane bikini with white donut glaze.” Really absolutely horrific and disgusting things that are these workarounds to get it to create sexualized images.

This has been happening for a long time online. There’s always been, since these AI tools have come out, problems with nonconsensual imagery being generated. There are lots of so-called “nudify” apps, right, that take regular dressed photos of people and undress them. And there are communities that share these as revenge porn and use them to harass and intimidate women and all kinds of vulnerable people. And this has been a problem.

People are trying to figure out the right ways to put guardrails up to stop this—to make sure that these communities get shut down, that they don’t continue to prompt these bots to do this, trying to get these tools to stop doing this. And a lot of this has been happening in these small backwater parts of the internet, and it does bubble up to the surface. But what’s changed here with X and Grok is that Grok is, as I said earlier: It’s baked into the platform. And so what has essentially happened is that X—xAI, Elon Musk—they have created a distribution method, and linked it with a creation method, and basically allowed for the viral distribution of these nonconsensual sexual images. And it has become, in the way that it does in places like 4chan and other backwater parts of the internet, it’s become a meme in this community. And people have decided that they are going to intimidate people and generate these images out in public.

And so what you have is publications posting photos of celebrities, and then a bunch of people, you know, in the comments saying: “@Grok undress this person.” “@Grok, put them into bikini.” “@Grok, put them in a swastika bikini.” “@Grok, put them in a swastika bikini doing a Roman salute.” And then you have a photo of a celebrity, undressed without their consent, in a Nazi uniform, giving a Nazi salute.

This is stuff that I have seen all across the platform. Not going into strange backwater areas of it—just looking directly at it. So this is out there. Something I noticed earlier this week—we’re recording this on Wednesday—was there was a photo of the Swedish deputy prime minister at a podium, giving a talk. And a bunch of people were asking Grok, prompting Grok to put her in a bikini, et cetera.

X and the people who work there have issued a statement saying that they’re working on the guardrails for this system. This is against their community standards, and they will punish the people who are involved here. But that doesn’t really seem to be happening. Just yesterday I was looking around, and people who are asking Grok to put women in compromising photos have blue checks next to their name, which means they asked the company for a verified badge. Those people are still on the platform as of this time when I’m talking to you.

So I reached out to Nikita Bier on his personal email. He’s the head of product at X. And I asked as a journalist, as a human: How someone can in good conscience work for a company that’s willing to tolerate this type of thing? Like, what’s the rationale? Who’s being served? How can you tolerate your product doing this? Do you imagine you’ll be able to get this under control with the appropriate guardrails? And if not, how can you sign your name to this stuff? How is this allowed to be in the world? They did not respond. They forwarded me to their comms lead, and I asked the same questions of them, and they never responded back to me. I have also asked Apple and Google similar questions. How can they allow an app like this on their app store? And they also have not gotten back to me.

The lack of response to this from the people who are the stewards of this platform, and the people who can exert pressure on this—including X employees or investors, or Elon Musk himself, who has made jokes about the @Grok bikini-photo stuff on the platform over the past week.

The lack of apologizing. The lack of urgency in trying to fix this. The lack of really seeming, from my perspective, to care about this, I think, feels a bit like crossing some kind of Rubicon. This is not a standard content-moderation issue. This is not a bunch of people trying to scold for something that is a part of some kind of ideology. This is basic human decency: That we shouldn’t have tools that can very easily create viral content of women and children being undressed against their will. Feels like the lowest possible bar, and yet the silence is—it just speaks volumes of what these platforms have become and what their stewards seem to think.

I would just ask of truly anyone who works at these platforms: How do you sleep at night with this? The silence from X, from employees there who we’ve tried to contact just to get some basic understanding of what they’re doing and how this can be allowed. And what’s happening on the platform, because the platform is not taking enough action to stop this, because it’s still allowing this undressing meme to go forward. What’s happened is: A culture has evolved here. And that culture is one of harassment and intimidation. And it feels like the people who are doing this know that no one’s going to stop them. They’re doing this out in the open. They’re doing it proudly. They’re doing it gleefully.

Something has to change here. I’ve been covering these platforms for 15-plus years, and I’ve watched different people in these platforms struggle with moderation issues in good faith, in bad faith. I’ve watched it devolve into this idea of politics and ideology. I’ve watched people pledge to do things, and then give up on those things.

It ebbs, it flows. The internet is chaos. I get it. But this is just different. This is a standard of human decency and social fabric and civic integrity that you can’t—you can’t punt on it. You either choose to have rules and order of some kind of a very base level, or you just—it does become full anarchy and chaos. And it seems to be that’s the direction where they want to go.

So if you work at X, if you’re an investor, if you’re somebody who can exert any influence in this situation, I would a) love to hear from you. And also I would ask: Is this okay? Is this what you want the legacy to be? Sorry for getting on a soapbox there, but I think it’s a massive, massive story. And one that, again—I think if this is allowed to just be the way that the internet is, then we lose something pretty fundamental.

So anyhow, it’s a tough way to segue there, but today’s conversation is actually the opposite of all of this. I do a lot of tech criticism. Do a lot of really sort of, you know, aggressive reporting, trying to hold tech companies to account. And that means looking at a lot of awful things and talking about a lot of awful things. But today’s podcast is about something great, something that’s actually hopeful that’s being built. It’s about a group of technologists who’ve come together with a different vision for the internet: a positive vision for the internet, something that they are trying to build that can sort of lead to positive outcomes and people living their best lives.

And so this project is called the “Resonant Computing Manifesto.” Basic top-line idea of it is that technology should bring out the best in humanity. It should be something that allows people to flourish. And they have five core principles here, that are essentially meant to combat the hyperscalers and extraction of what we know as the current algorithmic internet that we all live on.

And to talk about that, I’ve brought on Zoe Weinberg, Mike Masnick, Alex Komoroske. They are three of the writers of the “Resonant Computing Manifesto.” And I had them on to talk about why they came up with all this, and what, if anything, we can do to change the internet in 2026.

Warzel: All right: Zoe, Alex, Mike. Welcome to Galaxy Brain.

Mike Masnick: Thanks for having us.

Warzel: You all put forward something that I actually came across very recently. Often my timeline is a mess of the horrors of the world. The terrible things, the doomscroll. And this kind of stopped me in my tracks, because frankly, it wasn’t doomscrolly at all.

And when I clicked on it, I began to feel this very strange emotion I’m not used to feeling, which is hope. And/or, I agree. And I agree, and it doesn’t make me furious. And so what you guys have done in part, with a group of other people, is come up with something called the “Resonant Computing Manifesto.”

And it is based off of this idea of resonance. And when you guys put this out—and I want you guys to describe all of this—but when you put it out, you said that you were hoping this was going to be the beginning of a conversation. A process about getting people to realize technology should work for us, and not just for the people at the very top, the people behind [Donald] Trump on the inauguration dais, that sort of thing.

And so, in this world of mergers and acquisitions and also artificial intelligence and all that jazz, I wanted to start the conversation off with a definition of what resonant technology is and what it means. And I’ll bring that up to either all of you or one of you.

But what is resonant technology? What does it mean to you?

Alex Komoroske: So to me, that resonant computing. There’s a difference between things that are hollow—leave you feeling regret. And things that are resonant—leave you feeling nourished. And they’re superficially very similar in the moment. And it’s not until afterwards, or until you think through it, or let it kind of diffuse through you, that you realize the difference between the two.

And I think that technology amplifies whatever you apply it to. And now with large language models that are taking what tech can do and making it go even further than before, it’s more important than ever before to make sure the stuff that we’re applying technology and computing to is resonant.

And I think we are so used to not having a word for this. And we can tell that something is off in slop or things that are just outrage bait or what have you. And social networks, but we don’t know how to describe it. And just having a term for that: the kind of stuff that you like.

And then also the more that you think about it, the closer you look, the more you like it. Does that capture it?

Masnick: Yeah, pretty much. I mean, we spent a lot of time trying to come up with the term.

Komoroske: And you wanted something that was ownable, that was distinctive, that wasn’t just a thing that would fade into nothing.

Zoe Weinberg: There’s a lot of terms out there that now have a lot of baggage. Even something that sounds kind of innocuous—responsible tech—I think now comes laden for a lot of people with a bunch of associations or different movements of people, whether it’s corporate or grassroots or otherwise.

And so, you know, we were trying to move beyond that a little bit in the choice of the word resonance.

Warzel: Yeah. There is also like—there’s an onomatopoeia to it. There’s sort of, this is what it sounds like. You have resonance there. And also there is something a little bit, the word that comes to mind is almost monkish.

Like, a monastery type. There’s something that’s very, it’s … resonance is not, like, a capitalistic word. It is a word that signifies something much different to me. Like sort of sacred. You know?

Komoroske: Yeah. It’s balance, pureness. tThere’s something about it that feels very whole, maybe.

Warzel: And at the top of the manifesto, there’s this line that is sort of offset there. A pull quote, if you will. Says: “There’s a feeling you get in the presence of beautiful buildings, bustling courtyards. A sense that these spaces are inviting you to slow down, deepen your attention, and be a bit more human. What if our software could do the same?”

That was the thing that struck me there. When did you guys see a sort of architectural element to this? Like, an inspiration from things that we see and experience in meatspace, so to speak, in the world?

Komoroske: We had the word resonance, I think, actually came before we …so, I’m a big fan of Christopher Alexander. He lived a few blocks away from me. And, you know, a big fan of The Timeless Way of Building and a few other books.

And so we had various formulations of it, that try to key off of that frame or idea. I don’t think he ever calls it resonance in the book, in his actual book. But, you know, it’s a word that other people—maybe he might offer it as one of the potential names. He calls it aliveness and wholeness and other things.

But so, it was always in the mix of the kind of vibe that we were trying to capture. And then we decided to lean into resonance and introduce it via this architectural lens. And actually, that addition at the top was a late addition, because it starts off talking about resonance kind of indirectly and it pivots into this architectural frame.

And someone was like, What? I thought you were talking about technology. We said, Okay, let’s put a little teaser about the architectural connection up at the top, to help connect with the way the middle of it is going, so you don’t get confused.

Weinberg: I think there’s something also powerful about writing and thinking about software, which exists in a digital plane—that is, not a physical space—that feels like it’s kind of in the ether and a little bit untouchable. And then trying to ground that in a very human reality, which is in fact tied to place and space and where we spend time.

And maybe drawing some insights from those physical realities into the way in which we build digital spaces.

Komoroske: Christopher Alexander, when you read some of his work and, we all know that feeling, we can all imagine the situations that we’ve been in, the environments where we feel that resonance. And there’s something very, I don’t think we ever think about it in the digital world. Because you have to be, when you’re in it, in the physical world. It’s impossible to ignore it when when you’re in it.

And there’s always point. Let’s—why don’t we ask the question, why do digital experiences not feel the same way? They absolutely could. You know.

Weinberg: And I think, you know, what is the feng shui for software? It’s maybe a way of thinking about it. But I think that goes much deeper than UX and UI-design principles.

It’s much more about: What is the experience as a user, and as a human, interacting with a tool over repeated periods of time?

Warzel: Well, and I think too, a lot of—at least what I reach for in my work, which a lot of it is, critiques of, you know, big-tech platforms and such. A long time ago, I found the word architecture—the “architecture of these platforms”—as just being extremely helpful to communicate some of this stuff.

I think there is a way for people who, you know, are just using these platforms to get from A to B. Or, you know, on the toilet at a moment of just, I’ve just gotta get away from the kids, or whatever it is. If you’re not thinking with the critical lens—which, there’s no judgment there—about these platforms, you might just sort of think this is a neutral thing. Or this is a thing that just does a thing, and, you know, whatever. And I think that, you know, architecture—this idea that there are designs, there is an intentionality to this algorithm or this layout or whatever choice that a platform has made that leads to these outcomes—that leads you to post more incendiary things, or whatnot. And I think that architecture there is so helpful to let people see like: no, no, no. In the same way that, you know, these arches are the way they are. This stained-glass window does this, to give this vibe. So is putting the “What are you thinking?” bar right here, or whatever. The poke icon wherever.

Komoroske: So I think that’s also about, with connection to architecture, that’s even stronger there. I think of traditionally, architecture is, like, this designed-top-down cathedral. Like, the designer’s intent. And one of the things that Christopher Alexander later did was this bottoms-up emergent of: How is this space actually used and modified? How does it come alive?

And I think that’s one of the reasons architecture, in his sense, I think really nails it. Because a lot of these experiences, like a bunch of people, when you build Facebook 10 years ago, were trying to connect the world. That’s a prosocial outcome. It’s prosocial in the first order. The second-order implications, turns out, oh, actually are not prosocial.

And so you get these emergent characteristics that are not what anyone intended going in, necessarily. And still, and yet, they emerge out of the actual usage of how different people react off each other, and how the incentives kind of bounce off each other. And so I think architecture hits that emergent case too.

Warzel: Mm-hmm. So, Mike, I’ll throw this to you. How did this come about? What is the behind-the-scenes process here? I’ve heard, you know, “We’re using these words, and we’re taking ’em for a spin in the world for two weeks.” This does not sound like something that you guys wrote last weekend and put up on the thing.

There’s a lot of people behind it who aren’t on this call here. Or this podcast here, I should say, not a call. How did this come about?

Masnick: Yeah, I mean a lot of this is Alex, and so I’m curious about his version of this. But in my case: I mean, I met Alex about a year ago. Almost exactly a year ago at some event.

And we got to talking, and it was a good conversation. It was a resonant conversation, where I sort of came out of it saying, Oh wow, there are people thinking through these things and having interesting conversations. And then we kept talking and he said, “You know, I’ve been having this same conversation with a group of different people. And I thought I might just pull them all together, and we’ll get into a Signal chat, and we’ll have a Google Meet call every couple weeks. And we will try to figure out what do we all—we’re all having this feeling, what do we do about it?”

And then we did that for almost a year. I mean, it’s kind of incredible. And where we would just sort of be chatting in the group chat and occasionally having a call and sort of talking through these ideas, and working on it. And trying to figure out even what we were going to do with it.

Weinberg: I definitely think the manifesto emerged very organically.

Masnick: Yes.

Weinberg: To the point that I would say in the first couple months of us meeting, Charlie, like I was like: Okay. It’s really fun chit-chatting with these interesting people that Alex has brought together, but let’s get to brass tacks. Is this going anywhere? And I have to say, there was a part of me that wanted to end those calls being: Okay guys, what’s our agenda? Where are we going? What are the outputs? How are they met? Whatever. And I actually think, Alex, you did a really great job of kind of keeping people from jumping to that sort of action-item mode too early.

And so, from my perspective, we did not get together to write a manifesto. We got together to talk about these issues. And then, very naturally, you know, out of those conversations came a set of ideas and principles, and sort of theses. That then felt like we should put them out in the world.

Warzel: Did this feel like—the choice of the word manifesto and the choice to just do this—does this feel a little bit, too, like a response to we’re in a manifesto-heavy moment here? It feels like there are a lot. Whether we’re talking like the Marc Andreessens of the world or, if you pay taxes in San Francisco, you need to write a manifesto to get your garbage picked up or something.

But is this a response in the same way? Or is it meant to be seen as, in some senses, in dialogue with some of these other things that are coming out there?

Komoroske: I think to some degree, I don’t know, actually. I can’t remember how we ever discussed if it should be a manifesto.

We just knew that there should be something that we could point people at, that kind of distilled some of the conversations and ideas that we were having. And I think I’ve seen a bunch of manifestos in the tech industry, that sometimes I look at and go, Oh my God, is that the tech industry that I’m a part of?

That doesn’t seem at all like—that seems so cynical or so close-minded about the sort of broader humanistic impacts that technology might have. And so, I think the choice of doing something that other people have, you know … this manifesto was deliberately kind of humble. It says: We don’t have all the answers; just here’s a few questions that seem relevant to us.

That was a very important stylistic choice. Manifestos are not typically humble. But we aimed for that because we wanted to almost counter-position to some of the ones that say, This is definitely the right way. And everyone should think about it this way.

Masnick: Yeah, I almost think I’ve been using that as a joke to other people. Where it’s, This is the most humble manifesto you’ll ever see.

Which is not something—you know, you don’t normally see those two words together. You don’t think of as a manifesto as being humble. But, I mean, this was definitely a part of the conversation that we had. Which is: We want to be explicit that we don’t have all the answers, and that this is the start of a conversation. Not, you know, putting an exclamation point on a philosophy or something.

Weinberg: I do think, Charlie, you’re touching on something noteworthy here. Which is, and I’ll speak only for myself, but I’ve been observing in the last couple years as it has felt to me like the ideological landscape of the discussion in Silicon Valley has been really defined by these extremes.

And on one end, it’s like the accelerationist kind of techno-optimism way of seeing the world. And on the other side, on the other kind of far extreme, it is like existential and catastrophic risk and ways that, you know, we must prevent that. And I know a lot of people who don’t feel like they really belong in either of those camps, and actually don’t even really think that the optimist/pessimist spectrum is like the right way to think about it.

And so from my own perspective, part of what I have hoped that the “Resonant Computing Manifesto” will accomplish is, like, helping to establish some values and some north stars that are kind of on a different plane from that conversation. That also feels like there can be both. You can both be optimistic about the ways things might develop, and also concerned about the places we’ve come from. And that those things can coexist, and that is like the beauty and complexity of the technological moment we’re in.

Masnick: Yeah, totally. Because, you know, I had written something in response to Andreessen’s manifesto, and I never really thought of this as like a response.

Warzel: Is it the “build one” or the “techno-optimist” manifesto?

Masnick: There’s been many. Yeah, that’s true. Fair enough. But, you know, I’ve always considered myself, and I’ve been accused of being, a techno optimist. Like, to a fault. And like, I am optimistic about technology. But to me, his manifesto really, you know, rubbed me the wrong way. Because I was like, This isn’t optimism. What he was presenting was not an optimistic viewpoint.

It was a very dystopian, very scary viewpoint. And so soon after it came out, I had written a response, like, “That’s not optimism that you’re talking about.” And there, and if you really believe in this—this vision of like a good, better world from technology—then you should also be willing to recognize the challenges that come with that.

Because if you don’t acknowledge that, and don’t seek to—if we’re building these new technologies—understand what kinds of damages and harms they might create, then the end result is inevitably going to be worse. Because something terrible is going to happen. And then, you know, the politicians will come in and make everything else that you want to do impossible.

It’s just like: Think this through. Like a couple steps ahead.

Komoroske: And so technology is powerful. Like, we should be careful with that power, and we should use it for good. And I think it is incumbent—you know, it’s a good thing for people to do, to use technology for good. Like, you shouldn’t sit there and not use it.

You should use it, and you should be aware of the second-order implications and the third-order implications. And not say, “Well, who could have seen this inevitable outcome?” You know, so much in the tech industry is about optimizing. It’s about driving the number up. It’s about thinking, not necessarily thinking about second-order implications.

I, at some point, had somebody tell me, You know, anything that can’t be understood via computer science is either unknowable or unimportant. Which is an idea that, you know, pervades some parts of Silicon Valley. And I think this combination of the humanistic side and the technology side into the synthesis, I think is where a lot of value for society is created. And you have to have them in balance. You have to be in conversation with each other.

Warzel: Well, that’s definitely speaking my language, for sure. That’s like Charlie bait right here. But I want to define a little of this. I want to actually define it, but first I want to define it via its opposite.

What’s the opposite of resonance here? How would you describe the current software dynamic? I’ll let anyone who wants to take that. But maybe all of you, honestly.

Komoroske: And to me it’s just, I think, most of the technology, the tech experience and consumer world is hollow. In that you wake up the next day and go, God, why did I do that?

Or you use the thing. To me, if you use a tool and then after you are sober, after you’ve sort of come down from it, because sometimes you’ll be really hopped up on the thing. So maybe a week later, or the next day, would you proudly recommend it to somebody you care about? And if not, then it’s probably not resonant.

And you know, at some point, somebody—I was having this debate with somebody at Meta many years ago—they said, Oh, Alex, despite what people say, our numbers are very clear. People love doomscrolling. It’s like, that’s not love. That’s right. Like that’s a … what are you talking about?

So I think trying to just make number go up, and increase engagement or what have you, is what creates hollow experiences. And that tends to happen when you have, hypercentralized, hyperscale products. One of the reasons that happens inevitably is if you have five hyperscale products that are all consumer, and trying to get as many minutes of your waking day, there’s only so many waking minutes of people’s time in a given day. And so you naturally kind of have to marginally push. You know, try to figure out the thing that’s going to be more engaging than the other thing. And that emerges, I think, fundamentally when you have these hyperscale products—which is what emerges when you have massive centralization.

And all these things are of a piece, and lead to these. Hollow. Yeah.

Masnick: I think there’s a concept that has come up a few times in the conversations, in the various meetings that we had. And I don’t remember if it originated from you, Alex, or from someone else. But like, the difference between what you want and what you want to want, which may take a second. You think through, and you begin to like, Oh, right.

Like, there is this belief within certain companies that revealed preference is law. “If people love doomscrolling, ’cause they keep doing it, then we’re just giving them what they want.” Like, shut up. Like, you know, anyone who complains about that is just wrong. But then, as Alex said, it leaves you feeling terrible.

You have a hangover from it later. Whereas, if there’s this intentionality—of like, No, this is what I really want; I get nourishment out of it; I get value out of it in a real way—that lives on. That stays with me; that lingers. That’s different. And there’s that intentionality. As opposed to like, the problem with Oh, people love to doomscroll.

It’s, yeah. Because you’re sort of manipulating people into it. And people feel that they might not be able to explain it clearly. But like, it just feels like someone’s twisting the knobs behind the scenes, and I have no control over it. Right. And I think that feeling is what pervades; it’s the opposite of resonant computing.

Weinberg: I also think the opposite can be defined as any technology that’s ultimately undermining human agency. And so that can be things that are attention, you know, engagement-maximizing. And so it removes your agency in that sense. ’Cause you’re not actually able to express what you really want.

But also all the kind of micro ways in which we end up feeling deeply surveilled by the technology that we use. And I think all of us have probably had moments where we feel deeply creeped out by our tools. And I think, to me, that is the opposite of resonance also. So part of it’s about attention and engagement. And then part of it also is about, you know, having some individual autonomy in how you make decisions, where your data lives, who has access to it. And all of that we’ve tried to kind of embed into this piece.

Warzel: So you all write in the manifesto—and I’m going to quote you guys here, back to you at length. Hopefully it’s not cringey because it is written, you know, with a committee of people; I hate when people read my own stuff back to me.

But you all say: “For decades, technology has required standardized solutions to complex human problems. In order to scale software, you have to build for the average user, sanding away the edge cases. In many ways, this is why our digital world has come to resemble the sterile, deadening architecture that Alexander”—mentioned by you guys before—“has spent his career pushing back against. This is where AI provides a missing puzzle piece. Software can respond fluidly to the context and particularity of each human at scale. One size fits all is no longer a technological or economic necessity.”

This is the one part where I was tripped up while reading, and not in the “I am reflexively against AI” kind of way. But because personalization, I feel, in my own experience a lot of times, can be discordant with that idea of resonance.

I think personalization can be great. I think it’s actually, you know, underutilized or -realized in the tech space. But when I look around at the algorithmic world that we’re living in, sometimes it can feel like optimization. Which was, you know, the word there—like personalization and optimization comingle together.

Yeah. To become part of the problem and not the solution. So I was curious how you all would respond or think about that.

Komoroske: I think that the key thing there, I agree with that. What is the angle of the thing that is personalizing itself for you? Is it the tool, is it like trying to figure out how to fit exactly into the crevices of your brain? To get you to do something that is … you know, to click the ads or whatever?

Or does it feel like an outgrowth of your agency? Like, one way I talked about it is: Large language models can write infinite software. They can write little bits of software on demand, which has the potential to revolutionize what software can do for humanity. Today, software feels like a thing.

You go to the big-box store, and you pick which one of the three beige boxes, all of which suck, are you going to purchase. And instead, what if software felt like something that grew in your own personal garden? It was something that nourished you and felt like it aligned with your interest naturally and intrinsically, because it was an extension of your agency and intention?

And I think that kind of personalization—where it doesn’t feel like something else manipulating you, but it feels like something that is an extension of you and your agency and intention—I think is a very different kind of thing. We’re just not familiar with that kind, because it doesn’t exist currently.

Warzel: I was going to ask Alex just to push—not push back on it, but further follow up on that. Is there anything that exists like that, you think? A piece of software that feels garden grown versus a big-box store?

Komoroske: The one that keeps me coming back in my history is like—I think looking back at the early days of the web, actually, is where you had a bunch of these interesting bottoms-up kind of things.

HyperCard is my favorite one from many, many, many years ago. Have you heard of HyperCard? It’s like this thing that allowed you to make little stacks of cards. And you could have images on them; you could click between them, and you could program them to be like slideshows. Or like stacks of different things. And interlink.

The original game Myst, that was really popular, was actually implemented as a HyperCard stack back in the day. And so HyperCard, to me, is an example of one of these tools that allows you a freeform thing, that allows you to create this situated, very personalized software. You could argue that spreadsheets also have this kind of dynamic, because it’s an open substrate that allows you to express lots of different logic and build up very complex worlds inside of itself.

It’s pretty intimidating, but it is something that gives you that kind of ability to create meaning and behavior, inside of that substrate.

Masnick: Yeah. The thing I’ll say, to that point—and you’re not the only one who has sort of stopped on that line. And a few people have called it out and raised questions about it. And I think it’s because the idea of personalization, to date, has generally really been optimization. And it’s been optimization for the company’s interest, as opposed to the user’s interest. I think the real personalization is when it’s directly in your interest—and it’s doing something for you and not the company.

In the end, it has to be the user who has the agency, who has the control. Who says, This is what I want; this is what I want to see. And having it match that.

Komoroske: Charlie, I’ve also made a bunch of little tools. You know, a bunch of—if you’re technical, you can build these little bespoke bits of software now that fit perfectly to your workflow with large language models.

And that’s the kind of thing that a few of us can see a glimpse of this today, who are at the forefront and able to use Claude code and in the terminal to make these things. And I think in the not-too-distant future, large language models would, put on the proper substrate, will allow basically everyone on Earth to have that same kind of experience, that feels like an extension of their agency.

And I think that’s what some of us are seeing. And that’s why it’s in that essay. And that people who haven’t seen that yet are like, Excuse me, what? Like, you know, because they haven’t experienced it yet, they can’t see what’s coming.

Weinberg: Yeah; I do think that that sentence itself in many ways is a little bit forward looking. And so, as Alex said, there’s glimpses of it.

But I think the urgency and feeling like we needed to write about this is that it feels, I think to many of us, like the introduction of AI into all of our workflows gives us this kind of amazing opportunity. And crossroads. To either build along the lines of the paradigm of big tech and platforms and everything we’ve seen in the last, you know, couple decades—or we can try to shift into this new paradigm that is about personalization that, as Mike said, is not extrinsic from a third party, But something that you are building intrinsically yourself.

Warzel: I want to go through, actually, some of these starting principles. You all have five of them

that are these guiding lights. And I’d love to just sort of rapid-fire go through them, have whoever wants to explain just a little bit about how you’re thinking of them. Or how they, you know, might work to give a framework or a set of ethics or values to whatever is going to come out of this manifesto.

Right. And how they could be incorporated. And so the first one here is “private.” Which it says: In the, in the era of AI, whoever controls the context holds the power. Data often involves multiple stakeholders, and people in the service stewards of their own context, determining how it’s used.

We’ve talked a little around that. What “private” makes me think of, in a world of AI, is like: Our consumer-AI tools look the way that they do now because they’re built by the people who have spent—not totally, but when you think about like X, Google, Meta—the people who have spent the last, you know, 10, 15, 20 years collecting information on people.

So you are going to build a product that makes having that information more valuable to the end use. That’s part of the architecture there. But talk to me about how you see that first principle. Yeah. Zoe, do you want to take that one?

Weinberg: We debated this word a lot, and even the concept of privacy.

Komoroske: Yeah. We debated all these words.

Weinberg: Yeah, that’s true. But, you know, I think this one in particular is tricky, because we really went back and forth on—is it privacy that we feel like is the key value here? Or is it really about control, and putting the user in the driver’s seat?

And so it’s about, you know, consent. Rather than it is about just, like—and I think I speak for all of us. Like, I don’t think any of us are privacy maximalist. There are lots of, you know, amazing, wonderful prosocial reasons that you don’t always want to keep information private. And actually sharing information can be very helpful. And all those things.

And so, I guess, there’s a different way that we could have framed this that was a little bit more about control, or about agency, or whatever. But I think there is something meaningful about privacy as a value, and the notion. And the point of having privacy in the digital world is to be able to have a rich interior life. And that is, in many ways, very central to the experience of being human. And that’s why privacy is an individual value. It’s also a societal value. And I think that that was sort of important to capture in the mix here.

Komoroske: What we try to do with all these words is the word themselves.

We want to communicate on its own. And, if anything, go a little bit too hard in the direction it’s going. And because we actually soften the statement a fair bit about data stewardship. Because, you know, various thoughtful people pointed out that, well, actually data is owned, co-owned by the different parties. And in some cases you do want to give it up for an advantage, and whatever.

Mm-hmm. But we wanted the word to be private. Like, we wanted to be obvious when you have these five words. Like you could apply it to a product and say, “Does this fit, or does this not?” And not have little, like, soft, nuanced words for some of this. So we try to add the nuance in the sentence after the key word.

Warzel: Well, to that point, Alex: “dedicated.” You guys define it as: “Software should work exclusively for you, ensuring contextual integrity where data use aligns with expectations. You must be able to trust that there are no hidden agendas, conflicting interests.” Why’d you use the word dedicated? Like what do you mean exactly?

Komoroske: I wanted something that was, again, about: It’s an extension of your agency. It is not a conflict of interest, because it is in your interest. And “contextual integrity” actually is a meaningful phrase, because this is Helen Nissenbaum’s concept of contextual integrity. Which is, to my mind, the gold standard of what people mean when they think of privacy.

And it means: Your data is being used in line with your interests and expectations. So it’s aligned. It’s not being used against you, and it’s being used in ways that you understand or could be—or would not be surprised by if you were to understand it. And so that we wanted to get the words contextual integrity in there to get across this alignment with your interests and expectations.

Masnick: I think that’s a really important concept. You know, one of the discussions that comes up when talking about privacy is this idea that privacy is like a thing. And to me it’s always been a set of trade-offs. And the thing that really seems to upset people is when their data is being used in ways that they don’t understand, for purposes that they don’t understand.

And that is the world that we often live in, in the digital context. It’s like we know we’re giving up some data for some benefit, and neither side of that is fully understood by the users. We don’t know quite how much data we’re giving up. And we’re not quite sure for what purpose. And we’re getting some benefit, but we can’t judge whether or not that trade-off is worth it.

Warzel: I think about this all the time in terms of the “terms of service” agreement. I try to tell people, with that: Imagine that on the other side of the button that you were about to click is the most expensive-looking boardroom that you’ve ever seen in your life. With a whole bunch of people who make more in a week than you do in a year.

All in fancy suits. You know, like perfectly coiffed. And they’re just standing there, being like you versus them, you know? That’s what that is. It’s not a fair fight. You are agreeing to things. Yeah. Anyway, I want to keep running through this, though, because I want to get to ask a couple more questions here.

But the third of the five principles is “plural.” Which is: No single entity should control that distributed power. Interoperability. That seems relatively obvious. But, I mean, is this the idea of the decentralized, Bluesky sort of, you know, protocol-type thing? Being able to port your information to that, just being like a central tenant?

Masnick: Obviously that’s a big tenant for me.

Warzel: Yeah, I was going to say, I spent a lot. And you were involved with Bluesky, correct?

Masnick: Yes, yes. I’m on the board of Bluesky. But I also wrote the “Protocols not Platforms” paper that sort of was part of the inspiration for Bluesky. So, that kind of thinking—I’ve spent a lot of time thinking about that thing. And so I did. But I do think it’s important, not just in the social context—it’s important across the board. And this idea of, you know, why I’ve always thought that Bluesky or just a protocol-decentralized system is so important is this idea that we want to avoid giant centralized systems that will continually manipulate things. And so making sure that we don’t go down that path with, you know, the AI systems, I think, is really important. And just putting out there the idea that now, at this stage of the development of AI, we should be thinking about that. Rather than what we’re doing with social. Which is having to go back, you know, a decade: Oh crap. Like, Oh wait, we shouldn’t have done that.

And it’s funny to talk to the early Twitter people, who were like, Yeah, you know, we kind of thought that’s what we were doing. And we just lost track of it.

Warzel: Well, and it’s also like the biggest form of competition, actually, Like if you have a place where you can just say, I mean, I feel like I’m seeing this. I’ve seen this so much with the newsletter game.

Yes. Like, you have a lot of people who came to a company like Substack, just because, okay, yeah, this works really well. Great recommendation system. I can grow this audience; I can do this, I can link it to my, you know, paid. Boom. Like it just works. And then some of those people have problems with the leadership, the direction of the company, whatever.

And because of the way that, you know, newsletter lists work, and things like that. And the portability via, you know, different payment companies. You can just, you know, pop it over, and it’s relatively seamless. And then, of course, you have companies trying in ways to, you know, lock in these ways to keep people.

But this idea of interoperability: is that like competition? It allows Ghost or Beehiiv to, yeah.

Komoroske: Plurality is one of the things that leads to it. Also, it’s important to make sure you don’t have that undue influence of one particular voice. It’s important also to have competition and adaptability. A healthy system has multiple options that are multiple opposite, who are trying and competing to be the best version of it. And if we all used a single model, for example, and we didn’t realize what its bias was, or what it else could do, that would be bad. And that’s one of the reasons that having most people using just a single chatbot of ChatGPT, which obviously only works with OpenAI models, is not nearly as good of a feature as one where people can use different models in different contexts and try them out and switch between them.

Warzel: The fourth principle here is “adaptable.” Anyone can take it. It does seem relatively understandable.

Komoroske: The way I think about that one is: It lifts you up. It doesn’t box you in. ’Cause a lot of products have, like, if a product manager said, These are the five actions you were allowed to do in this context. I want a system that’s open-ended, that I can use to build whatever I want to do. As opposed to something that kind of limits me into a particular subset of things that I can do.

Warzel: And the last one is like—this is my music, man. “Prosocial.” Technology should enable connection, coordination; help us become better neighbors, collaborators, stewards of shared spaces, online and off.

This dovetails—we can talk about that all day. I’d love to hear what you guys think about it. I went through some of the comments of people who are, you know, who are seeing this. Who want to be either signatories or contributors or just help out with the process. A lot of really interesting comments.

A lot of people writing their own thoughts. One of them kind of hit me a little. I guess resonated with me a little bit. And it was the culture. It was, I’ll quote them: “The cultural backlash against attention extraction is coming. Technologies that respect and protect human attention will, in time, win the marketplace.”

To this idea of, like, the prosocial: I think it’s pretty obvious that these tools are having antisocial effects. Not always; not in every context. But there are, you know, ways they’re trapping us, keeping us from living the lives we want to live. In some contexts, making us feel just bad, or adding to problems with mental health that people may be having.

I’m curious about this idea of the cultural backlash, though. Zoe, I’d love for you to begin on this one. But do you all feel like this is happening? To me, it feels very much like people are waking up to the idea that, like, This stuff makes me feel bad, and I don’t know how much longer I really want to feel bad in this context.

Weinberg: You know, it’s funny; I exist in this world of tech and start-ups and VC, where everybody is really excited about AI and thinks it’s really positive. But if you take even a half a step outside that bubble, I think it is very clear, at least to me, that the AI backlash is coming, or it is already at our doorstep. Or it’s already here. And that there is a lot of hate and vitriol, and I get it. Because, I think, Charlie, you nailed it. Like fundamentally, I think what people are reacting to is that AI, in many ways, has been profoundly antisocial.

Even in the ways that social media itself were bad, it’s almost like it’s almost gotten worse. Like, I’ll give you an example. Like, we used to worry about people falling down these, you know, these sort of disinformation rabbit holes, because they’re in these echo chambers on social media. Now, you can fall down a disinformation echo chamber, a rabbit alone with a chatbot. You know, it’s like an echo chamber of one.

Warzel: Made it real simple.

Weinberg: It’s the way that I think about it. And that’s even more antisocial than the previous version, which, you know, what was itself very problematic and very, very harmful. And so I think that’s part of what people are reacting against.

And, look, I live in New York City. There was a subway campaign for a product called friend.com that elicited a ton of backlash from the city. And, you know, I’ve been observing things like that—and a few other instances along the way—that have definitely convinced me that I think for most people, whether or not they’ve used AI tools or they feel like AI is coming for their job or not, there’s just this sort of instinct of like, No, I don’t want this in my life.

Komoroske: Especially as an extension of the tech of the last decade. I mean, this industry, the one who gave us this crap and this hypercentralized. People who make these bombastic statements that are unnuanced and just don’t really seem to grapple with the amount of power and responsibility that they have.

Like, that’s not the place that you want AI to be. I also think, by the way, there’s a difference between AI tools. AI should not be your friend. If you think that AI is your friend, you are on the wrong track. AI should be a tool. It should be an extension of your agency. The fact that the first manifestation of large language models in a product happens to be a chatbot that pretends to be a human … it’s like the aliens in Contact who, you know, present themselves as her grandparents or whatever, so that she can make sense of it.

It’s like—it’s just a weird thing. Perfect crime. I think we’re going to look back on it and think of chatbots as an embarrassing party trick. You know, in five years, and be like, Oh, that was the wrong manifestation of large language models. Large language models should be in this inherent tool thing where you don’t get confused about whether this thing is your friend and you don’t get, you know, caught up in delusions of grandeur and everything.

Warzel: Well, I think too with the backlash, like—you mentioned this with this idea. It’s like, Oh, these companies are going to build, you know, the next generation of it. But I think, too, that it’s bigger than AI. Like I think you see this a lot.

This is, I think, the third time I’ve said this on this podcast now. But you can feel with younger generations that they understand very acutely how they’re being manipulated. Like they’re born into this ecosystem that a lot of people have had to take time to learn and understand.

There’s this real idea of it. And there’s sort of—though they are a part of it in a big way, also really, they don’t suffer fools in that sense. It’s like, I don’t necessarily want that. I’m feeling bad about it. It does feel like when I want to get hopeful about this stuff, I talk myself into this idea that we are sort of on the cusp of a little bit of a change.

I’ve experienced in the last year more phone-free spaces in general. Yeah, yeah. Right. Like this idea of this thing is not helping me in context, you know, outside of where I want to use it as a tool. I’m going to put it away right now. Or I need, I need someone to create a permission structure for me to put it away.

Komoroske: Going to saunas, I’ve heard, is a big thing, because you can’t have phones in them. Like, as a social space to be in person.

Weinberg: I predict that in the next year, we’re going to start to see people creating human-only spaces and saying, like, Okay, just so you know: This gathering, whether it’s online or in person, this is a human-only space.

Like, no wearables. Like, don’t bring your AI assistant or your, you know, Copilot. I’ll go on record. That’s a prediction for 2026.

Masnick: I was gonna say, I think one of the interesting things is that society adapts to these things. And there is this belief that like, oh, you know, once we start spiraling down, we continue to go down. But, like, people and society as a whole starts to figure this stuff out. And it may take a while, and there may be a lot of damage done in the interim.

Today, going on Facebook, I mean—they’ve picked other places. You know, over time as new generations come in, they sort of look at the old stuff, and they realize, they see the problems of it. Because, you know, they’re all much more obvious. And then they look for somewhere other space.

And so, you know, in the social world—that had been TikTok, for example, which has its own problems. But if there’s going to be another generation, and there’ll be another generation, of AI tools. And there’ll be another generation of social as well. And if we’re in a position where we’re creating spaces that are welcoming, human people will move to them eventually, as they realize how problematic the other ones are.

And that’s, like, a lot of the response that I’ve heard—at least to the manifesto as it came out. Was like just this—like an exhale. Like, yes. Like I’ve been thinking that we need this, you know, vision. And I’ve been thinking about it. I didn’t realize other people were thinking it. And I think that’s part of society, you know—moving forward with these things and thinking through. Like, what is next?

What do I want? If I’m going to make a jump to new tools and new systems, you know, like I want to be a little more deliberate about it. And if the people building it are also more deliberate about it, maybe we can actually have a next generation that meet these principles that we’re talking about

Warzel: To that end. there are some interesting critiques here that are made, I think, in good faith. One of them, I wanted to just highlight and get your reaction to, which is: Somebody on Bluesky said, quote, “Like other cyber libertarian frameworks, they stop short of the root cause, which is politics. Liberation depends on shifting political power, because power determines which values take hold.” That’s obviously true. I think other criticisms that I have seen in general that seem to be, you know, part of the cynicism of living in 2025—or 2026, as people listen to this—is this idea that it’s like, Yeah, that sounds great, you know, in theory.

But again, you butt up against the politics of it all. The capitalism of it all, the scale of it all. Like all of all of those things. Very real. Yep. By the time people hear this, like Warner Brothers may be bought by like 450 companies.

We don’t know what that future looks like, but all of them portend some kind of strange dystopian consolidation. No, but yeah. But in general, like, how are you guys thinking about that? This is a guiding statement, to some degree. This is not meant to, you know, solve every problem that exists.

But how are you thinking about that? But, you know, coming up against politics of it.

Komoroske: I think the one broader point—by the way, I published another essay about optimization and how modern society just kinda optimizes everything. It’s true in the technology industry, but it’s also true in tech, or, sorry, in business and in politics too. I think it’s the defining characteristic of modern society—that we forgot that optimization actually does come in at a cost.

It’s just an indirect and harder-to-see cost. and I think that that is true across many different dimensions. It’s part of what I think everyone is feeling in this moment. And I think I would also point out that we are part of the industry. And we are also realists. We understand the incentive structures, the things that get us stuck in these kinds of behaviors.

A couple things. One, I think a lot of this is to a point that we made earlier in the conversation. Some of it is totally structural, and it’s, you know, the person at the top making these kind of decisions to optimize for Wall Street or something. Other parts of it are just emergent. They’re just local product managers in a given team making a decision: Okay, oh, we know that number is supposed to go up.

And it’s not thinking about what the downside of number going up is. And actually, if they think about it in terms of resonance, they might make a better product that actually creates more value for the shareholders, too. It doesn’t have to be intention. So little things, if everybody can say, Hey, is this resonant?

Just having people be able to have that terminology, and ask that question. If lots of different people are asking that throughout the industry, that could have an impact. And second, myself and a number of others who are working on this manifesto are working on things that are structural changes. To the kinds of distribution structures and power structures that create technology.

I’m working on an alternate security model that’s open and decentralized and allows getting rid of some of these silos that lead to aggregation, while still being fully aligned with people’s private interests. And so, we are not just saying: “Oh, what if everyone just said, Hey, let’s be nice today?”

You know, there’s some of that that actually could be somewhat effective. And also, we are realists about the emergent factors that cause some of these things. And working to modify or tweak or do what we can to help, you know, the right kinds of things emerge.

Masnick: There are many reasons to be cynical right now.

I completely understand where all that is coming from. And I think some of the job that we’re hoping to do—or at least I’m hoping; I shouldn’t speak for anyone else on this—is like, the more that we can paint this picture and show people. And yes, like, maybe some of us are a few steps into the future on this stuff. but if we can start to bring that back, begin to show people there are real things behind this, and we can all start to make decisions in this direction. And hopefully we can start to thaw out some of that cynicism and show that there’s something real here.

And each one of those steps is important. We’re not going to, you know, flip the entire structure of the world right now. But we can take these little steps and really make a difference over time.

Weinberg: The only thing I would add is that I think that there’s already been a lot of ink spilled doing the diagnosis. And I think capitalism is part of it. I think our political system is part of it. I think optimization culture is part of it. I think it’s a confluence of different factors. But I think part of what we were trying to do, at least in this piece, is move beyond just the diagnosis of the problem and try to craft a positive vision for where we should go.

But absolutely: A totally valid critique might be that you need to spend more time unpacking some of those underlying drivers. Of which we are all, I think, very aware of the ways in which that shapes, you know, the current reality.

Warzel: So I want to land this plane with: People are going to be listening to this at the beginning of the year.

I think this is a hopeful vision of a future. Or at least telling people. What if you planted the seed in your brain of a hopeful vision while you’re constructing these things? What is giving you all hope about what’s coming next this year in this space?

You’ve gone through this. You are clearly hopeful people to put this together in some sense. No matter how, you know, beaten down and cynical anyone who exists online is these days. But yeah. What is keeping you guys going forward on this?

Weinberg: I think what gives me hope on this vision is that I am seeing this whole new generation of founders and technologists, many of whom are like contemporaries of me, that grew up kind of under big tech and are just questioning all of the assumptions that underlie the way that we built things. And are trying to think about building things in new ways, and I think are very subscribed to the types of values and vision that we lay out in the manifesto.

And so I think that’s what gives me hope. I feel like the tide is really turning. And the fact that there’s been a ton of interest and momentum in the manifesto itself, I think, suggests to me like, you know—there’s a critical mass here who feels this way? And that’s kind of all you need to, like, nudge it in the right direction, I think.

Masnick: Yeah, I was going to say, like, I guess I’m the old man of the crew. I think that I’ve been alive slightly longer than the others. And that I remember the early days when people were thrilled with new technology, and it was exciting. And before it all seemed to turn. And to me, there is this element of going back to that. You know, there are mistakes that were made, but being able to go back to that time while recognizing the mistakes and doing a better job this time, I think, is actually really important.

And I’ve had, like, some of the criticism I’ve seen. Because I talked about this concept of going back, and some of the criticism was like, No; it was always terrible. And it’s like, No, like, I lived that time. And I remember when using new technology in the internet was enjoyable and exciting. And we can bring that back.

There’s nothing that says we have to keep the awful parts of the internet working the way that they currently work. And really against our own interests. And so I’m very optimistic; when you put these things out in the world, you know, people are gravitating to it. And that’s the first step toward pretty massive change over time.

Komoroske: I think for me—I think people have felt so cynical, and that they can’t do anything. And like, maybe they’re the only one that wants to push back against some of these optimization pressures. And seeing the response that people have to this has been really inspiring to me. Because at some degree, I’m thinking that we’re saying this thing that no one’s going to care about.

Everyone’s going to think it’s kind of dumb. And people are like: Yeah, how can I participate? I’m like, Oh my gosh; wow. Okay. I mean, I’m into it too, but so it feels very encouraging to me to see people feel that agency and wanting to sort of change the world in this way. And again, I work with a bunch of folks who are at the cutting edge of using large language models, and interesting ways to create infinite bits of situated software that—you know, personalized software.

And like, it’s exciting what you can do with some of these things. And again, I think chatbots—if you’re looking at chatbots like this is going to be social media, but worse and just kind of the same old story of centralization. Like, my hope is that we will be beyond that relatively soon as people start waking up to all the other things that you can do that are now possible. And democratized and available to just about anyone, to aid and empower them.

It’s really cool. And so I’m just extremely excited about what we as a society are going to do with some of these technologies.

Warzel: All right, with that, let’s go forth into 2026 and make it suck less than it did before. No, I appreciate everyone’s time. Zoe, Alex, Mike: Thank you for coming on Galaxy Brain and offering a unusual dose of positivity and hope.

Masnick: Excellent. Well, thanks for having us. Thanks.

Warzel: Thank you again to Zoe Weinberg, Mike Masnick, Alex Komoroski. I wanted to have this conversation because back in November at this panel discussion that I participated in, in Bozeman, Montana, we had this long conversation about the generative-AI moment. And so much of it was focused on the economic issues, the fears of artificial general intelligence, the ways in which this is all being abused. The conversation—as it tends to with new technologies that are consequential—it gets very negative, and very reactive, and very thinking about all the scary externalities of a new technology.

And at the very end of the conversation, one of the panelists, Sarah Myers West, who does a lot of work in AI policy, ended with something that was very—to borrow the term—resonant to me. And that was that she was really tired of talking about all the bad stuff and all the stuff that AI shouldn’t be—you know, the future that is being brought to the world that we need to fear—and wanted to think about ways to put forward a positive vision. To stop being on the defensive all the time and to think about: What is the future you want to build? If this technology is here, if it’s not going away, how do we harness it to do something that will be productive and helpful to human flourishing? And that just stuck with me, especially as someone who’s always focused on these negatives. And so a couple days later, when I saw this manifesto, I just thought to myself, Some of this stuff is probably idealistic. Some of this stuff is gonna be really hard to enact from.

From a political standpoint, from a fundraising standpoint, it’s gonna be a challenge. It’s always a challenge to build something that resists scale in general. But that doesn’t mean that we shouldn’t try. We shouldn’t try to be so rational about all of this, that we talk ourselves out of building something that matters, that helps, that actually aligns with the goals of being a good human living a good life. And so I found the conversation—in that sense more than anything—to just be motivating, to be something that, as we continue to do episodes here, as I continue to do my reporting, as you all continue to live your life out there among this technology, to think about. What it is you want. What it is we should be building to come up with positive visions of how this stuff should work, instead of constantly just defending against it.

So I hope this conversation gave you some of those ideas, some of those tools. It certainly did for me. And it’s something we’re gonna be continuing to explore throughout the year. So thank you once again. If you liked what you saw here, new episodes of Galaxy Brain are dropping every Friday.

And you can subscribe to The Atlantic’s YouTube channel, or you can go on Apple or Spotify or wherever you get your podcasts. Please leave a five-star review if you would. And just remember, if you also enjoyed this, you can support this work and the work of all of my colleagues at The Atlantic by subscribing to the publication at TheAtlantic.com/Listener. That’s TheAtlantic.com/Listener. Thank you so much for listening, and I’ll see you on the internet.

Everything Reacting to Everything, All at Once

2026-01-06T18:21:41-05:00

This weekend’s attack on Venezuela produced plenty of indelible images. The one burned into my brain was shared by President Donald Trump on Truth Social. Defense Secretary Pete Hegseth is sitting in front of a laptop at a makeshift command center in Mar-a-Lago. He’s monitoring the raid with a grave expression on his face, eyes intently focused on something out of frame.

At first glance, the image has all the trappings of a Serious Tactical Raid Photo, à la Pete Souza’s famous Situation Room snapshot, which showed President Barack Obama and his national-security team tracking the raid on Osama bin Laden’s compound. But then you see what’s behind Hegseth: a large screen displaying an X feed. The photo is blurry, but it seems to show Hegseth and company using X’s search function to monitor tweets about the raid. On the screen, hovering over Hegseth’s left shoulder, is a giant face-holding-back-tears emoji (🥹).

The photo quickly spread around the internet on Saturday—mostly as a way to mock just how terminally online the Trump administration appears to be. “They monitor the situation just like how we do,” one person who works in crypto wrote on X. On Bluesky, I watched others make fun of Hegseth, Trump, and Secretary of State Marco Rubio as part of a “podcaster-occupied government.”

It is no secret that the Trump administration is social media–addled. Over the past year, most of the government’s major online accounts—especially on X—have become megaphones for cruel and racist shitposting, not unlike what one might see from a garden-variety troll on 4chan. These accounts have shared deportation ASMR; an AI-generated, Studio Ghiblified version of a real photo of a crying woman being arrested by ICE; a post comparing immigrants to the alien vermin in the Halo video-game series; and Nazi-coded “Defend the fatherland” memes. And who could forget the AI-slop video of Trump in a fighter jet dropping what appeared to be human feces on protesters in Times Square. These official government communications are a key part of how the Trump administration does its job. It is governance through content creation.

This is why the Trump administration is staffed with former reality-show stars, cable-news hosts, and popular podcasters. It is why the government allows friendly camera crews to accompany ICE raids, why former Representative Matt Gaetz is given a Pentagon press credential along with Laura Loomer, and why Vice President J. D. Vance spends his days trolling people on X. It’s why Kristi Noem staged a photo op in front of a cage full of men at El Salvador’s Terrorism Confinement Center, and why the administration has allowed YouTubers to make videos there. It’s why an assistant attorney general at the U.S. Department of Justice is on X, grousing about her follower count stalling at a paltry 1.3 million and asking, “What kind of content do my folks want to see more of to like and share?” And it’s the reason that Katie Miller—who left the Trump administration to work for Elon Musk and then left Musk to start a podcast—posted a photo on X on Saturday showing a map of Greenland colored as an American flag with the caption “SOON.” The U.S. government is concerned first and foremost with spectacle, engaging in both fan service for its most extreme supporters and the constant trolling of its enemies. The goal, above all else, is to elicit a response.

The Hegseth photo from Saturday and others like it confirm this dynamic. Why would the men in charge of the most powerful military and intelligence services in the world be monitoring a popular X account called “OSINTdefender” if they weren’t performing for an online audience? (The White House did not respond to a request for comment on the matter.) Perhaps one could defend their scrolling as potentially useful data gathering on the grounds that, during the bin Laden raid, early tweets from local citizens began to break the news well before the mainstream media caught on. But such posts are low-level intelligence, a kind that is arguably far too trivial for a Cabinet secretary or president to pay attention to. The simplest explanation is that everyone in that room at Mar-a-Lago wanted to observe the spectacle created by their actions.

“I watched it literally like I was watching a television show,” Trump said in a phone interview Saturday with Fox News. “It was an amazing thing.” His description suggests military invasion as personal entertainment—a reality show with stark geopolitical consequences that the president can produce and direct via his whims.

Trump is obsessed with ratings, and social media provides ample opportunities to watch numbers go up at the same time as other politicians, media members, and onlookers respond. And so you get not only an invasion and a press conference but a slew of posts. In the first day after Nicolás Maduro was seized, Trump or official government accounts had shared:

High-resolution war-room photos
Footage of the invasion set—without any shred of irony—to the Vietnam protest song “Fortunate Son”
A meme stating, in bold red and white lettering, “Don’t Play Games With President Trump”
An angry photo of the president beneath block lettering of the acronym “FAFO”
A video mash-up of Trump, Rubio, and Maduro set to Biggie Smalls’s “Hypnotize”

The more I watched the fallout from the invasion play out online, the more futile any effort to make sense of it felt. I felt like I was trapped in a recursive loop of a very specific style of internet content: the reaction video.

Reaction videos began on YouTube in the mid-2000s, and they are now a hallmark of online content. As the name suggests, they show people responding to other media. In the early days, these were typically gross-out videos like the infamous “2 Girls, 1 Cup,” and as The New York Times’ Sam Anderson wrote back in 2011, their appeal was allowing “people to watch this taboo thing by proxy, to experience its dangerous thrill without having to encounter it directly.”

Over time, reaction videos became far more interesting and varied—a way to experience something vicariously for the first time or to share the joy of something you love with others. As Anderson noted, the videos are best at capturing surprise: “that moment when the world breaks, when it violates or exceeds its basic duties and forces someone to undergo some kind of dramatic shift.”

Today, the world feels like it’s breaking in any number of ways—so perhaps it makes sense that the logic and structure of the reaction video pervade media, culture, and politics. Some of this feeling has to do with the structure of social media, where timelines are no longer sorted chronologically but algorithmically, feeding users a steady stream of content that’s likely to elicit a strong reaction. The algorithmic internet has always been chaotic, but as the platforms have matured and evolved, the culture they produce and behaviors they provoke have become insular and inscrutable—at least to people who don’t spend huge chunks of time online. Especially on X, algorithmic culture is characterized by ceaseless iteration: Everything that’s happening is piled atop all the things that just happened.

When any given event occurs—a raid in Venezuela, say—the trolls, pundits, know-it-alls, and shitposters flood in immediately. Discourse is a gameable phenomenon now; people know how to play their roles by heart. These days, one doesn’t experience the news on these platforms before seeing the memes and reactions—the reaction and the news are, in essence, one thing now.

By the time I saw the news of the raid, a photo Trump had posted showing a blindfolded Maduro in a Nike sweatsuit had already become a meme. Just a few minutes later, that meme had mixed with a dozen others. In a few hours, I stumbled upon a split-screen generative-AI slop video of Maduro DJing in the sweatsuit in one frame while, in the other, Trump’s face was superimposed onto a clip of Jon Hamm blissed out and dancing in a club. An image of Maduro in handcuffs wearing a blue sweatshirt and giving a thumbs-up was followed immediately by a marketing meme posted by that sweatshirt’s manufacturer; scroll more and there’s the same photo, now with Maduro’s face replaced by Charlie Kirk’s.

The result is essentially insane and postliterate. But it is also pretty much legible for those steeped in online culture. It is coherent incoherence, everything reacting to everything else, all at once. The same thing happened after Kirk was shot. The memes, commentary, and speculation became a culture unto itself, a loop of ironic posting, information warring, and commentary on commentary—all before his shooter was identified or Kirk was even pronounced dead. This process is nihilistic, and it has a dehumanizing effect. Stories about people or countries in conflict become abstract, buried under a pile of memes and recursive references that exist for little more than scroll-by entertainment. Over the past decade, online performance for others has evolved out of popular culture and media and become a primary means of communication for everyone—mass shooters, meme makers, and POTUS all included.

Trump has been rightfully called the Twitter president in the past, and a crucial part of that legacy is the skilled exploitation of this information environment. His administration’s chief output is online shitposting. It’s not an actual form of governance, nor is it a kind of policy, but it is performative speech that’s supposed to signify action and, in the case of the Venezuela raid, strength. The resources of the most powerful military in the world are being marshaled in service of making memes declaring, “THIS IS OUR HEMISPHERE.” All because the country’s leaders think it’s good theater, and in a postliterate political era, the spectacle is propulsive. It gives so many of the entities of our media, political, and cultural ecosystems what they crave: something to react to.

How About a Little Less Screen Time for the Grown-Ups

2025-12-26T13:00:00-05:00

Subscribe here: Apple Podcasts | Spotify | YouTube

Are your parents addicted to their phone? In this episode of Galaxy Brain, Charlie Warzel explores how technology is affecting an older generation of adults. Instead of a phone-based childhood, Warzel suggests, we may be witnessing the emergence of a phone-based retirement—one shaped by isolation, algorithmic feeds, and platforms never designed with aging users in mind.

To untangle whether this is a genuine crisis or a misplaced moral panic, Warzel speaks with Ipsit Vahia, chief of geriatric psychiatry at Mass General Brigham’s McLean Hospital in Massachusetts and a leading researcher on technology and aging. Vahia emphasizes that older adults are anything but a single category, and that screen use can be both protective and harmful, depending on context. The key, Vahia argues, is resisting reflexive judgment. Ultimately, this is an issue not of screens versus humans, but of how families navigate connection in a world where attention is mediated by devices in every age group.

The following is a transcript of the episode:

Ipsit Vahia: Don’t go, You’re spending too much time on the phone. Instead, perhaps ask, What are you watching on your phone? What apps are you into? This is what I do with my phone. You could use their phone use as a conversation starter, as a way to meet them where they are, as a way to perhaps enter their world rather than expecting them to jump straight into your world. And, you know, it can just be the basis of strengthening connection rather than breaking it.

Charlie Warzel: I am Charlie Warzel, and this is Galaxy Brain. About a year ago, around the holidays, I began to hear a similar complaint. People were heading home, often with their kids in tow, to be with family. It was there that they noticed that their parents, or grandparents, or older relatives were behaving differently.

Broadly, the complaint was that their older loved ones seemed consumed by their devices—constantly on TikTok or Instagram or Facebook, watching vertical-reel videos. Sometimes they said they found it hard to hold a conversation. In multiple instances, people reported that some of these adults seemed to not pay much attention to their grandchildren.

Most of the people that I spoke to recognized it pretty quickly. It was the same thing they’d seen in their own kids: a screen-time problem. So, naturally I was curious. I wanted to get a sense of the scale of this. So I asked around on social media. I got dozens of responses over the year. From young people, from older people. Lots, lots of people.

Some older folks, they wrote in to tell me that they felt bad about how much time they were beginning to spend on social media. Others told me they’d found joy in the process and that there was no problem and I was over-hyping it. But many confirmed the anecdotes. Some feared that their loved ones were growing depressed or anxious as a result of a problematic relationship with their screens.

Others worried about older relatives falling victim to scams. Almost all of them, though, stressed that this felt like an emergent phenomenon—something that had popped up since the pandemic. I heard stories like this one from Josh.

Josh: It’s super interesting to watch my kids and my dad interact in the same space. With my kids, they love screens. They’ll spend an hour most mornings watching Bluey or Sesame Street or something. But when it’s off, they generally switch gears. They’ll go bike, they’ll do gymnastics, they’ll play board games. They engage with the world around them. My dad, on the other hand, is constantly glued to his screen.

He’s reading the news; he’s scrolling through his email. With my dad, there is no off switch. When we look at photos from his trips to see us, they show the kids engaging with their grandma, playing games, being silly, while grandpa’s in the background playing a game on his iPad.

Warzel: Or this one from Kim.

Kim: I’m 55. I have tween twin girls. I worry a lot and spend a lot of time controlling their screen time. And it’s kind of a joke, because if they saw the amount of screen time that I have in a day, it is way more.

Warzel: Kyle worries about what his parents are seeing.

Kyle: It’s really tricky to talk to my parents about anything news-related.

My parents are both, you know—they’re very intelligent, they’re thoughtful people. But media literacy is a problem for them in a way that it isn’t for my teenage kids who were kind of raised with an understanding of the dynamics of digital content. I mean, we all spend our days staring at screens. But the screens that my parents are staring at is this really toxic combination of Facebook and Fox News.

So it gives them these distorted views of things. You know, like: Portland is violent; New York City is super dangerous; immigrants are selling fentanyl to schoolkids; isn’t [Zohran] Mamdani anti-Semitic? You know, that kind of thing. And it’s hard to break through that information bubble. I’ll call my mom out sometimes for sharing disinformation online. But like, how do you tell your mom she’s participating in a Russian disinformation campaign? I sound like the crazy person in that conversation.

Warzel: But perhaps the most affecting one came from a nurse in the United Kingdom

who told me what she sees in her ward.

Nurse: I’m a nurse in the U.K., working in an inpatient ward. Most of our patients are in the 50-plus age group, and the majority have smartphones or iPads. When you’re stuck as a patient in the hospital, a lot of the time you’re bored or lonely or both. That can mean loads of really excessive screen time. It’s probably the 50-to-75 age group I’m most worried about, because they’re tech savvy enough to be where they want to be online, but they’re not necessarily media literate.

They might not recognize harms or understand how algorithms funnel consumption in certain directions. Some of it is fairly benign, like being obsessed with fake-AI animal stuff or compilation videos of babies. And sometimes it’s actually been pretty funny, like when folk end up in an autoplay cul-de-sac of Chinese-language videos.

But I do think the negative effects of excessive scrolling are bleeding through more, mostly in the anti-immigration stuff we hear. And the conspiracy thinking, medical distrust too.

Warzel: These testimonies struck me in part because they sound quite a lot like the concerns voiced for years by parents about children and devices. In the last decade-plus, there have been endless panics—many warranted, and others less borne out by the evidence—about children and screens. That their young minds are being influenced or warped by devices designed to take advantage of them. In most cases, screen panics position children as defenseless, even agentless.

They’re confronting this force that’s powerful enough to cause problematic behaviors among their underdeveloped minds. But now it seems the problem exists on the opposite side of the age spectrum. Data suggest there’s a reason people might be noticing this more now, because more people are aging into a retirement era with more fluency with smartphones and tablets and social media. On YouTube, for example older people are among the platform’s fastest-growing demographic. It’s possible that the pandemic and the attendant isolation accelerated all this adoption, from rideshare apps to Zoom. The confluence here seems very real. Older individuals may have extra time, and they may be more socially isolated than other demographics—and they’re seeing their retirement era just collide with this extremely powerful algorithmic world of social networks, apps, on-demand streaming services, and even the arrival of generative AI.

These are things that confound people of all age groups. But older people are not by any means a monolith, and technological tools are very clearly lifelines for aging people. As well as tools that can bring great joy, information—help them live full and creative lives. This is a really complicated issue, and so I wanted to speak with an expert and find the perfect guest here.

Dr. Ipsit Vahia is the chief of geriatric psychiatry at Mass General’s McLean Hospital. He’s the director of its technology and aging laboratory, and he’s been studying this phenomenon—and more importantly, working with patients in clinical settings. He joins me now to talk about all of this.

Dr. Vahia, welcome to Galaxy Brain.

Vahia: Thank you for having me. Delighted to be here.

Warzel: So you head up the technology and aging laboratory at McLean Hospital. Can you tell me what you all do there?

Vahia: Sure. So it’s a clinical-research laboratory that’s focused on understanding the way older adults use technology, and then also leveraging technology in a clinical setting with older adults with dementia or other mental-health challenges.

So we have a broad range of areas in which we do research. This includes early diagnostics, technologies for monitoring and supporting clinical decision making. But they’re also developing interventions using tech.

Warzel: So how did you get into this line of work, especially working with people on the furthest side of the age spectrum there?

Vahia: There’s actually an origin story there. When I was a trainee, it was when smartphones first came about, and I think I remember this incident very specifically. It was the year 2009. I was a trainee in California, and my wife and I were out for dinner with friends, and we had a 4-year-old child in tow. And he was doing what 4-year-olds do. He was boisterous, and I saw a simple thing. Again—this is circa 2009, so this is quite common now.

But in 2009, I had never seen this before, where my friend took out his smartphone and gave it to his child. And the child was engaged with it, and we didn’t hear a peep from him. We made it through four courses of dinner. Glass of wine, even.

Warzel: Very common now.

Vahia: Now it’s default. But back then, the thing that it really made me think about was that: If this engagement with the screen could sort of stabilize the behavior of a child, could it do the same for someone that was, you know, functioning at the level of a child? Which is to say: someone with dementia. Could we use these devices to engage them? Could we use these devices to reduce agitation?

A little after that, when iPads came out, there was a different incident. So when I was working on the inpatient unit, we had a routine. And the routine would be that every morning started out with everyone gathering in the community area. And we would just read from the newspaper. And this was intended to sort of create the sense of community. A shared activity that brought everyone together. It also let us assess how people did in that group setting, because it’s a predictor of how they might do when they were on the outside. Now, on the morning that I was supposed to lead the meeting, the newspaper never showed up.

It was stolen, lost; we don’t know. But this was when iPads had just come out, and I happened to have a personal iPad with me. And an interesting thing happened that morning where, in the absence of the newspaper, I was able to pull out the newspaper’s website on the iPad. And we kind of went through the same exercise, but now it was digital.

And what happened was someone raised their hand and asked me—can you access only The San Diego Union-Tribune? I was training at UC San Diego, so that was the local paper. And I said, well, no, I can access any newspaper that has a website. Now, this was a Monday morning, and it was a very specific question. He said, “I’m from Pittsburgh. Can you tell me what they’re saying about the Steelers game last evening?” And so I did. I was able to pull up the column, and we talked about that when this happened. Another person raised their hand, and he’s like, “Well, that’s great. I’m from St. Louis. Can you find out what they’re saying about the Rams last evening?” And so I was able to do that.

And now, suddenly everyone was asking not for this one-size-fits-all newspaper reading, but they were able to get what was most important to them. And that was sort of the other big moment where I realized that you could, you know—with this device that we already had figured out engages people—we could also personalize the intervention.

And in many ways it was not about the tech at all. It was about what the tech made possible. And there’s a difference, because I think, to this day, some of the way we think about this is about the tech. But I’ve always thought about technology as a conduit to problem-solving and an intervention.

So as a clinician, the thing that we anchor our work around is: What is the patient need? Or, what is the clinical problem? And then think about—is the technology we have before us able to solve some of these?, And that served us well. I think that served us well.

Warzel: So tell me a little bit about—you work with this elderly population; you’re working on these types of interventions.

You’re also deeply attuned to the way that they use and interact with technology. Broadly speaking, how would you classify how people on this side of the age spectrum are using technology? Are they a monolith? Are they extremely different and varied? Like, how would you describe, you know, the elderly’s interactions with technology?

Vahia: Thanks for that question. I think that’s the question that really gets at the heart of it all. So, I think if our listeners learn exactly one thing from this entire podcast, it should be that. Older adults are probably the most heterogeneous group of all the age groups. And we don’t always think of it that way, right?

We think of the elderly as this one monolithic entity. I love your use of that word. And nothing could be farther from the truth. So if you pause and just think about this for a second. We think of everyone over 65 as part of this one block, right? We have infants, and then we have toddlers, and then we have pre-K kids, and then we have elementary school. And we are quite sophisticated in the way we compartmentalize people across the age span.

But then we get to age about 65, and they’re all seen as this one block.

So. In the “elderly” group, if we consider people in their 90s and people in their 60s, these people are 30 years apart. If you’ve seen and understood one older adult’s use of technology, you’ve really seen and understood one older adult’s use of technology.

Yeah, and I think this overgeneralization does not serve us well. Which is not to say that there is not truth in the data. I think older adults, as a whole, do use less technology, but it varies quite a bit by. age cohort. So, you know, 80-year-olds may not be quite as digitally literate around apps or mobile phones, but 60-year-olds assuredly are very proficient as a group. Now there’s exceptions, obviously, on both ends.

Warzel: Well, that makes a lot of sense. Right? You would expect that between a 65-year-old and a 95-year-old, there’s 30 years there—there’s a lot of life and context experience. And I think you’re right that we do paint people in a lot of age brackets, but especially the elderly, with this really broad brush.

But I am curious from what you are seeing—and this will contextualize a little bit of what I want to dig into in this conversation—but do you notice that there is a different effect on older generations in terms of the way that they are using technology than, say, younger generations?

Like, if you were taking the bucket of zero to 10 versus, let’s say, like 75 to 85. Do older generations use—like, is the effective technology different than what you see on younger generations?

Vahia: It is, again, with the understanding that one size does not fit all. Older adults as a whole—they’re slower to take up new technology, and they’re much more methodical about it. So I think older adults as a whole are less likely to just experiment or play with tech. They adopt technology when it serves a clear and defined purpose in their lives on the whole. So a great example was what we found during the COVID-19 pandemic and lockdown, right? Among older adults, being tech proficient actually predicted better mental health—and that’s because most of them use technology or newly adopted technology to stay connected. I’ll give you an example from our own work. We, like most health-care systems, sort of had this en-masse migration to telemedicine through Zoom or whatever. And we found that most of our patients were not already using this technology, and so we had to train them on how to use it.

And an interesting thing happened. We found that, of the people who were the majority that figured out how to use telemedicine through phones, et cetera, the ones who did best were the ones that learned Zoom. Not to keep their doctor’s appointments, but it was because their church started doing services virtually, or their family started having gatherings virtually. And then, once they learned it, they were using it way better and way more regularly and effectively than, say, younger populations. So the data are fascinating, because they find that high technology use in teenagers and adolescents is associated with worse mental health and is a predictor of sort of more isolation and loneliness, even depression. Whereas in older adults, engaging in technology seems to be protecting them from isolation and loneliness, and it seems to be enhancing connectivity. Now this finding might evolve over time, but broadly I think tech use and tech engagement is a positive for older adults, when broadly it’s more of a negative for younger adults.

Warzel: So that’s really fascinating and I think helpful in grounding what I want to get into here. Because there is a, I guess you’d call it like a meme online. But this is really just a whole bunch of anecdotal evidence that suggests that—I’ll put it this way—I have done a lot of reporting, talking to different people about elderly people and screen-time use. And a lot of what I’ve seen is these anecdotes from younger people. They go home for the holidays; they see that their loved ones who are older are kind of deeply engaged with their phones, with their iPads, with social media. In a way that younger people are recognizing—it’s potentially problematic. Or at least it makes them uncomfortable, right?

They come home; they say, I brought my kids over, you know, grandma and grandpa. Or Mom and Dad weren’t paying as much attention, right. They were just kind of stuck in their devices. This is really worrisome. And I have so many of these anecdotes that have piled up. Or you go on places like Reddit, and you see this “Help; my mom or dad has this screen-time problem.” And there is this developing feeling. I think you’re starting to see some news articles, and things like that, that say we associate screen-time problems with younger generations. We’re always worried about adolescents. And what they’re seeing—perhaps there is also this problem on the other side of the age spectrum.

I am curious: What is your reaction to all of that anecdotal evidence? Like, are you seeing this too? This idea that, where also technology may be beneficial, but are you seeing a screen-time problem forming generationally?

Vahia: So that is so interesting. Because I think the answer is yes. And I think we are seeing increased screen time among older adults as a whole. I think this is definitely true. But there’s a lot of nuance there. Because, I’ll preface it by saying that younger generations—you know, people are more similar to each other in the routines of their lives than not, right? Everyone goes to school. So there is sort of these like activities and routines that extend across the community.

And then, once we get older, you know, elementary-school kids are more like each other than middle-school kids. And middle-school kids are probably a little bit more like each other than high-school kids. And college kids are not quite as like each other. And then, you just continue to separate out. So as you get into late life, people have just had unique life experiences. And while there are similarities, I think there’s also a lot of differences in that life experience. And I think why that is relevant is—we have fewer sorts of ways to determine what constitutes problematic screen use.

So yes, there is increase in screen time; there is increase in screen use. But when that becomes problematic, you really kind of need to get into the weeds with each person to sort of decide if this is a good thing, if it is just what it is, or if it is a problem. To the example of people seeing their older loved ones at the holidays and finding out that they’re spending a lot more time with their phone than they used to. I hear that story in my clinic. I actually see that in my family—like, that’s probably familiar to a lot of people. And the way I think about it is, I mean, yes—you observe it when you meet them during the holidays. The problem is, you’re not there the rest of the time. And what are they doing with their lives the rest of their time? And is this a habit that formed because they just didn’t have all that much going on? And so, now their life is running more through their phone, through their perspective. Is it possible that they have a nice routine? Their phone is a big part of it, for better or worse. And your arrival is actually the disruption. Which is not—

Warzel: That’s so important though, right? Because there is this idea that you are dropping in to getting this window into their lives. Right? And when we talk about some of the issues, especially with people who are much older. Being isolated, being untethered from reality. Real life, right? Like, like civic life, right? If you can’t drive; if you’re in a rural or a remote location. And I think that’s a really helpful observation that this influx of people, or, you know, around the holidays or something is actually like aberrant is, is abnormal.

And the rest of the time these devices could be serving a really smart purpose. Or a really helpful purpose, rather. I wonder, though, when it comes to some of what is being seen—this is a separate part of this. It’s not just that when I hear these anecdotes that are reported that I am, or that they’re coming home and watching their loved ones be deeply embedded in their devices. A lot of the worry, too, is around what they are looking at, right?

This notion that they are scrolling on Facebook, you know, what me and my colleagues are calling “reel slop,” right? Like r-e-e-l. Where they’re seeing these AI-generated videos of things that are either, you know, misinforming them, or just strange and kind of detached from reality. And, like, really low quality, right?

Like, these aren’t the mitigations that you are talking about, where it’s allowing someone to play puzzle games that are sort of, you know, keeping their brain elastic. This is, like—kind of tuning everything out and just being washed over with low-quality slop content.

Is that a worry? This idea that the phones are helpful—and the connection is helpful and the tether is helpful—but what they’re seeing is potentially harmful, because it’s really low quality?

Vahia: It is a worry. And I think it’s real, and it’s consequential. So, the dark side to all of this screen news has a few different dimensions. I actually think the biggest one is that as older adults are spending more time on the phone, it’s getting easier for scammers to target them. And I think the screen-based scam targeting older adults—I think that is a real problem and a real threat. And with AI, it’s becoming even more sophisticated, because sometimes these scam tools can be really quite hard to distinguish from humans. Especially when the AI is talking to people. So I think that’s a risk. The slop is a risk too. Much has been said and written about misinformation in general, and older adults I think do tend to be a little bit more trusting of a technology that they adopt. I think that that innate skepticism isn’t always there.

And, again, the devil’s always in the details, right? If someone’s just scrolling through a social-media feed where they’re watching a video after the other, that’s a little bit different than two people forwarding content to each other. Or on a chat group, where there’s also communication and correspondence.

I think one of those things is—neither of them is great, but one of those is slightly better than the other. Because one of them involves interaction and communication, and the other one is just much more passive, which is less ideal.

Vahia: Correct.

Warzel: You know, participating in the same way, having sort of a panoply of options.

If you talk to younger people about their phones—and by younger people, I mean all the way up to, let’s say, 55, right? They’ll tend to complain about their use. They’ll talk about their doomscrolling, or I wanna get off this, or, It’s not helping me live my best life. But what are you hearing from older people that you meet with in terms of self-reporting? Are they worried about the time that they’re spending on their devices? Are they okay with it? How do people seem to feel about it?

Vahia: In preparation for this conversation, I kind of polled my colleagues. I work in a team with nine other aging and mental-health specialists. And I just told our team that, Have you seen this? Has anyone brought this up?

And the answer surprised me. That no one’s actually had any of their patients—and we see several hundred people—no one could really acknowledge or remember someone coming to them with problematic screen use as something to address. I think they were there for other things, and you sometimes uncovered a lot of screen use. But unlike, say, you know, substance use or alcohol use, or even things like gambling, we haven’t come across yet the issue of too much screen time as a bona fide problem that requires a mental-health professional. Others may have. So I think I’ll be watching the response to this to see if anyone can share a story. But we are seeing clear reports of more time being spent on the screen.

So, where my head’s at is—we are seeing people spending more time on their phone. But it’s not necessarily being thought of as a problem. And that’s interesting, isn’t it? Because if you’re spending way too much time doing something, you usually know when it’s a problem, versus when it’s not. And I see that as a signal that it’s probably got at least some benefits, or some positives.

Warzel: But do you think there’s also a literacy quality there? And what I mean by that is: Something I see from, especially, people in a younger generation than me—Gen Z, Gen Alpha—there is a real understanding, innately, having grown up around this technology, that they know they’re being manipulated at all times.

They know they’re being pushed by these algorithms into this thing. And there’s a frustration there, I think, because of just the understanding of the technology. It being so innate. Do you feel like maybe a little of this—maybe the lack of what you’re hearing on the end of the older people—comes from maybe not having that same media literacy? Understanding of the ways that the technologies work?

Vahia: I do think that that’s a part of it. But I also think it’s specific to this moment in time, and that digital literacy just takes time to trickle up the lifespan. So I think we are starting to see this shift. But these things are always going to start at the younger, more hyper-connected, more tech-literate generations, and then trickle up the age span.

There’s also the kinds of tech that older adults use. They tend to trust more mature, more subtle technologies rather than the latest, greatest thing. So, you know, most people are still happier about something like Facebook—which at this point counts as mature technology or at least a mature platform—and they’re less prone to whatever the newest ones are. Snapchat. We are paying attention to ChatGPT and sort of the new generative-AI models. I think a lot of people have their eyes on this, because every now and then we kinda see these leaps in tech adoption. So older adults historically were less prone to using computers. And by computers, I mean the classic desktops.

And then they were also—they used laptops a little bit more. But they were behind when cell phones emerged; they were not as quick to adopt cell phones. They were also slower to adopt smartphones. And then the tablets arrived, and that just seemed to mark this whole en masse onboarding of the technology because—it’s that Goldilocks phenomenon. iPads were just right. I think the screen was larger. The keys were larger. So just easier to type for people with sensory impairment or visual impairment. But also, they were so easy to use. You didn’t need to upload software; you didn’t need to download software. It was all kind of right there. You had to tap it. It was easy. So I think you see these generational leaps around ease and efficiency of use. And a lot of us believe that as these generative AI has gotten more—you know, as we’ve moved from typing to speaking, that’s marking a shift. It’s just so easy now, where you have a device and you tap in, and something is talking to you. And it talks back, and you can have a conversation.

So I think you have these like leaps every few generations of technology, and just simplicity of use. So I think we’re on the threshold of seeing a lot of change, as these voice-based AIs become commonplace.

Warzel: Are you seeing a lot of—just anecdotally—a lot of adoption of the voice-based AI?

Vahia: We are. We are. And, you know, it’s anxiety-provoking. Because I think it really brings all of the things that we’ve talked about to a head. That—I think it creates huge opportunities, but it also creates massive risks.

Warzel: Right. We recently had Kasmir Hill, a New York Times reporter, on the podcast, who’s done a lot of reporting around what people are informally calling “AI psychosis.” That’s not a medical definition, obviously, but this idea of problematic behaviors with chatbots. And something that she has noted, in the reporting that we talked a lot about, was this idea of the ways that these chatbots are so engaging, right? It’s not just that they mimic human nature and that they are conversing. Which I think—with someone who may be more isolated in general, or feeling like that—that is extremely attractive as a proposition.

But also this idea that they are prompting you to continue to engage, right? They are also sort of asking questions at the end of it. Wanting you to go further. And the more that people do engage, the higher the likelihood that you start to lose touch with what it is you are.

And this goes to people who are younger, too. This is happening sort of everywhere. The sooner you might lose touch with, Oh, I’m talking to a large language model, not a person, not a thing. Are you seeing any problematic examples of those interactions with chatbots? With some of the people that you’re seeing in the clinic?

Vahia: So personally, not yet. But it’s a matter of time. Because the thing that I’m nervous about is that bots—it’s that validation function. They rarely contradict during conversation. It’s more, it’s what you said. Like they’re designed to be facilitating, but they’re also designed to be validating. So a bot will not say no. A bot will say yes, but also if it wants to contradict.

And I think there’s a real risk there—that if someone has a question about something, and it’s risky. I’ll make up a ridiculous example. But, say, if an older adult were to ask their daughter, Should I send my bank-account information to this Nigerian prince? Their daughter would be, No. A bot might say, Well, that’s an interesting question. Here’s what you should know about this—that there is a scam like this, that maybe you should do this. Maybe you should do this. Maybe you should do that. And there’s a difference, qualitatively. Because one puts an end to a risky conversation, and the other may not put it quite as … sorry. One puts an end to the risky conversation, and the other may continue that conversation because it is designed to engage. And I think that is risky. Because that validation function, right? The bot rarely makes you feel bad by telling you you’re wrong. Even when it tells you you are wrong, it offers alternatives or other ways to continue the discussion.

Warzel: Well—and I think we, you know, should be clear here—or these purposes, that’s hypothetical. You know, it is possible these chatbots will—or that in some cases when you prompt them—will caution people against sending money to, you know, the theoretical Nigerian prince. But I get what you’re saying. Something you said earlier too, I think is very striking to this phenomenon. You know, I mentioned this, you know, short-form video-slop stuff that has historically been very prevalent on Facebook, and also Instagram. You mentioned that older people tend to adopt these more mature technologies, right? Like a Facebook. And I think what’s interesting as a technology reporter is that some of these younger, newer social platforms—they struggle with all kinds of emergent problems, but they’re also iterating out of them a little bit faster. Right? They’re sort of pushing the boundaries a little bit.

It’s interesting to me that you have these people who are on a platform like Facebook, that isn’t updating in the same way, right? Like it is happy to kind of keep that engagement. To not have those rules against, you know, these types of fake AI-slop images. And it feels, to me, like a danger that is not talked about enough potentially. That by not sort of evolving out of the platforms—like a Gen Z person might do—or being on the newest, latest, greatest thing that there is actually a little bit of this. Yeah, there is a danger of using an older platform that is not evolving in the same ways. Because then they get trapped with the lower-quality content. And I think that’s super fascinating.

Warzel: So one of the things that you’ve brought up here—that I think is one of the most salient points for people listening at home who may be dealing with an elderly relative or a loved one who they feel has a problematic relationship with some of their technology—is this idea that it can be really positive. That we should stop, pause, think about what role this is serving in their life.

You are in a clinic with people. You are using this technology in a way that is supposed to have positive interventions. Talk to me about some of the positives you’re seeing here with elders and technology use.

Vahia: So there’s many levels of it, right? The one thing I really try and emphasize is that you don’t have to always be using the most state-of-the-art, high-tech, fresh-off-the-lab tech.

There’s a strong case to be made for just teaching people to use well-established stuff. Properly. A very simple example is: I have people on our team whose job is to teach older adults how to use Uber and Lyft. Why? Because many of them don’t drive. Many of them are isolated. They’re used to calling a car service, or they’re used to calling for the ride. And of course these are benefits, not things they paid for. But, I mean, if I had a dollar for every time we showed someone how easy it is to call a car service that will take you anywhere. It can transform lives, food deliveries, and other examples.

So, you know, is it or is it not “technology” to teach someone how to use a widespread app? I would argue it is, because you are enhancing digital literacy, but you’re doing it around specific function. So some of it is just—people’s mood improves, people’s anxiety goes down. If you can simplify everyday functions that may be a challenge for them.

Warzel: What about in general? I mean, like, there’s those apps that help. But I think, you know, are you seeing positive effects with the social-media use?

Vahia: We can. But it depends on which social-media use. So, a big one is just text messaging, or things like WhatsApp or the Messenger apps. Why? Because if people’s social-media uses anchored around interaction and communication—rather than just the passive consumption of content—that’s a different thing. It’s sort of what I alluded to earlier. That in my family, I have people that—it’s actually quite specific to WhatsApp, that there are people on multiple WhatsApp groups just forwarding what you might consider slop. But it’s one thing to scroll by yourself in your room to watch slop. And it’s another thing to forward slop to each other. And then talk about that slop—whether it be “Is this real?” or “This is so stupid; what do you think?” So there is almost always value in interaction and communication.

I think in-person’s better … but in-person is not always an option, right? And so, you know, slop—when consumed in isolation—I think is almost universally a problem. Slop as giving people a common thing to talk about, that might not have too many common things to talk about? Now that’s a little more nuance, isn’t it?

That’s a little more positive. We know art therapy works. We know music therapy works. But very few people can play an instrument or draw. But if you give them an app that’s equalized artistic talent or musical skill, that’s a positive. So it’s not really about the tech; it’s about how you use it and how you apply it. And I think the art of digital medicine lies in that. The art of digital medicine, the art of digitally based psychiatry, the art of AI use lies in that. I’ll give you an example from an ongoing study, where we have a project where we are comparing a human geriatric-care manager versus an app that is trained on working with caregivers.

And this is all specific to dementia. Which is—it’s a very simple question. We generated, you know, a list of common caregiver questions. And we asked the same question to an AI chatbot and to a human geriatric-care manager. And then we did a third thing. We gave the human care manager access to the bot to see if they could come up with a hybrid answer.

And we compared differences.

But before we even get into what we found, the biggest finding was that it took our human six weeks to answer all the questions and compose their responses. It took the bot 13 minutes. And a lot of us sort of picked up on the fact that—even though we would not really question that you want a human resource, you want someone to help really work through whatever it is that ails you—the truth is, our human is not going to be available for a three-hour conversation at 11:30 in the night. AI is.

And AI is close enough to the … it’s not perfect, but there is something to be said for efficiency and access. I’m not saying it’s right. I’m saying you can’t discount it.

Warzel: All of that is so conflicting to me.

It’s right, because in one sense, I was kind of laughing earlier. Because of this notion of art therapy, music therapy … and then slop therapy, right? Like, sending it around to others and being connected. And I think that’s important, because it adds a rub to, you know, we look at somebody sort of canonically. There’s this … I don’t know if you’ve heard of Shrimp Jesus. Have you heard of Shrimp Jesus?

Vahia: I don’t think I’ve heard of Shrimp Jesus.

Warzel: Okay. It’s an AI-slop representation of this Christlike figure, but it’s a shrimp. And it was one of the early versions of AI slop that was very popular. And it seemed like … it was not fooling, but sort of bewildering, a lot of elderly Facebook users. Something like that. Anyway, those things are always presented as awful, right? That there’s somebody, they’re like brain rotting instead of generative in any way.

And I think that we have reflexively—especially someone like myself, a technology reporter—have classified something like slop as bad, right? You’re not gaining anything from it. And yet, what you’re asking people to consider is that, just as a meme—as a thing to trade back and forth, a building block of conversation, however silly it may be—or in general, if it’s fostering that kind of tether and that connection, I think that it’s important. And so that’s kind of confounding to think about. Something I wanted to ask you is: I feel like there is this idea that the technology is very helpful to people when it tethers them to reality, right?

Isolation. Loneliness. But I think what we’re also seeing, at the same time, is some of this tech, some of what they’re consuming is actually distancing them from reality. It’s blurring the lines of what is real. So you have this thing, it feels like two things are happening at once, right? Almost at the exact same time. Do you agree with that?

Vahia: I do. I do. And I think that conflict that you’re feeling, that confusion. That asking of, Well, which is it? Is it good or is it bad?

Warzel: I know.

Vahia: That’s actually the appropriate response, because nobody knows. But I think there are some guardrails to this, because the real answer is not, “Nobody knows.” But the real answer is, “It depends.” It depends on the person; it depends on the situation; it depends on the circumstance. I get asked all the time—you know, we now have therapy chatbots. And I get asked all the time—am I worried that these things are going to take away human jobs? And I don’t think so. In fact, I think it’s really sharpening the human effect. And I think it’s very close to what you said. That on the one hand, people value technology that tethers them to reality. But there’s also an untethering. And that’s exactly right, isn’t it? I think that the human function there is to then find the tethering, and to prevent that disconnection and that confusion.

And sometimes it’s as simple as acknowledging the confusion to begin with. We react poorly to ambiguity. I think there is this preference for clarity, and sometimes all we have to do is help people hold their ambiguity. But then do it while giving them some tools around how to then remain connected.

So: brain rot, slop. I think no one would argue … that’s probably not a good thing. But if brain-rot slop is giving you something to talk to people, preferably in the same room and face to face, and if you’re older? If it’s giving you something to laugh at, or something to at least make sure that everyone else is just as puzzled about it as you are? And then maybe it gives you an excuse to call up your grandchild and say, Well, what the hell is this thing? It makes no sense. Then, something positive has sprouted from that slop.

And, I think in many ways, I think there is a certain collective responsibility not to be absorbed by all of this—but to absorb it instead and assimilate AI as a piece that can promote. And this is all very Pollyanna. I’m not saying this is easy. I’m not saying this is how it’s going to go. This is messy, complicated stuff. But there is a reality where this can all be sort of leveraged into a collective positive.

Warzel: Yeah, my concern having covered this for a long time with the social platforms is that I think you’re right. And I would just want to say that I don’t want to paint with too broad a brush on this, and there could be these positive externalities from even the lowest-quality type of content. I think that’s something we all need to keep in mind. Where I worry—where I break a little bit from you is that these companies are generally very poor stewards of the regulations and the rules and the looking out for. And they do optimize for this engagement.

And if you have a segment of the population—be it 11-year-olds, or be it 84-year-olds who are showing signs of deeper and deeper engagement with a certain type of thing—the chances are it’s going to be fed to them at higher and higher rates. Right? And, that, to me is the concern. And that’s not on you, or that’s not on the people who are using this technology. That is, very simply, on people who are in charge of building and designing these platforms not serving their users properly. And that’s distinct from any kind of user behavior.

What I wanted to sort of end on here is: This episode’s going to come out during the holiday season. People are going to be at home. People are probably going to be experiencing this, we’ll call it a “phenomenon,” but just this experience of maybe seeing an older loved one immersed in a device. Maybe feeling a sense of concern. How do you suggest that people breach those conversations? And what should they be saying to someone if they do feel this way?

Vahia: Such a great question. I would say first—if you feel distress, see if you can hold it within you, and resist the temptation to jump to a conclusion about it. So don’t go, You’re spending too much time on the phone. Instead, perhaps ask, What are you watching on your phone?

What apps are you into? This is what I do with my phone. You could use their phone use as a conversation starter, as a way to meet them where they are, as a way to perhaps enter their world rather than expecting them to jump straight into your world. And, it can just be the basis of strengthening connection rather than breaking it.

But who among us responds well to being told whatever it is we are enjoying is wrong? Like, no one enjoys that. So, don’t do that if it bothers you. Fair game. But keep an open mind, and inquire and learn and assess what’s going on—rather than declaring it good or bad.

Warzel: I think it’s so smart that if we are talking about a behavior that seems to be isolating somebody, or seems to be drawing a human disconnect, that the appropriate way to respond to it is to connect with them, right? Not to disengage—or shame them in some way that may draw them further into their device, or further away from the loved ones in their life who they feel like they’re judging.

I think there’s something rather lovely about using this as an opportunity to foster the kind of connection that they may not be feeling. And that may be drawing them into that device.

Vahia: Yeah. It could be a reason to bond, rather than a reason to separate. Because we all bond over things we share in common. For better or worse, too much phone use is something we all share in common these days. Might as well use it.

Warzel: I think that’s a great place to end it. Dr. Vahia, thank you so much for coming on Galaxy Brain.

Vahia: Such a pleasure. Thank you for having me, Charlie, and for focusing on this. It matters.

Warzel: That’s it for us here. Thank you again to my guest, Dr. Vahia. If you liked what you saw here, new episodes of Galaxy Brain drop every Friday, and you can subscribe in The Atlantic’s YouTube page, or on Apple or Spotify, or wherever it is you get your podcasts. And if you enjoyed this, remember, you can support our work and the work of all the journalists at The Atlantic by subscribing to the publication at TheAtlantic.com/Listener.

That’s TheAtlantic.com/Listener. Thanks so much, and I’ll see you on the internet.

The Epstein Files Only Get Worse

2025-12-23T18:28:02-05:00

This particularly cursed holiday week kicked off in earnest last night when my father turned his iPad in my direction. On its screen was a terribly disturbing post on X containing two images. In the first, Jeffrey Epstein was hugging and kissing a little girl. In the second, that girl was bound and gagged on a bed.

Dad was rightly outraged and disgusted. He asked me if I’d seen the photos in my time going through the Epstein files. I immediately recognized the first image of Epstein and deduced that it had been Photoshopped from a widely distributed photo of Epstein hugging Ghislaine Maxwell. The second image seemed to be an AI rendering. (To add to the confusion, images reportedly do exist of Epstein cuddling children.) I let him know that the imagery was fake, and a distinctly non-yuletidy conversation ensued. Yes, Epstein was a heinous pedophile and convicted sex offender. Also, the internet is awash in fake, traumatizing slop that’s being used to score points in an ongoing information war. Happy holidays!

Early this morning, the Department of Justice released nearly 30,000 documents related to its investigations into Epstein. A previous batch was released late last Friday afternoon, as mandated by Congress, and was notable for its thorough redactions, its overall lack of material related to President Donald Trump, and the fact that it was incomplete. This latest batch contains far more mentions of Trump, leading the DOJ to issue a defensive-sounding, partisan, and frankly unprofessional post on X: “Some of these documents contain untrue and sensationalist claims made against President Trump that were submitted to the FBI right before the 2020 election. To be clear: the claims are unfounded and false, and if they had a shred of credibility, they certainly would have been weaponized against President Trump already.”

As I looked through the documents myself, I realized that many mentions of Trump in this batch come from news stories or documents referencing publicly available information about the president. For example, a random email in the archive includes a link to a story headlined “Trump: Kushner’s Security Clearance Is Up to Kelly.”

But there are some new, salacious-seeming details. Take, for instance, a 2020 email from an unidentified federal prosecutor alerting an unknown recipient that Trump had taken more trips on Epstein’s plane than was previously realized. There are at least two unvetted forms submitted in 2020 to the FBI’s National Threat Operations Center tip line that mention Trump’s name in conjunction with alarming and unproven allegations, including rape and paying for sex.

The White House did not respond to a request for comment about these new documents and allegations, and instead referred me to posts on X by the DOJ; Trump has previously denied any wrongdoing and has downplayed his past relationship with Epstein.

Another shocking revelation is a copy of a letter allegedly written by Epstein to Larry Nassar, a former U.S.-gymnastics-team doctor who was convicted of possessing child pornography, among other crimes, and who used his position to sexually abuse hundreds of women and girls. The letter was postmarked three days after Epstein’s death, in 2019, and makes a reference to suicide. “As you know by now, I have taken the ‘short route’ home,” the letter, which appears to have been signed by Epstein, reads. “Good luck! We shared one thing … our love & caring for young ladies and the hope they’d reach their full potential.” The letter continues: “Our president also shares our love of young, nubile girls. When a young beauty walked by he loved to ‘grab snatch,’ whereas we ended up snatching grub in the mess halls of the system.” The existence of a letter sent by Epstein to Nassar had been previously reported by the Associated Press, but the contents had not been; earlier today, the DOJ posted on X that it had concluded that the Nassar letter was fake, which “serves as a reminder that just because a document is released by the Department of Justice does not make the allegations or claims within the document factual.”

These details alone are a lot to take in. That they are just a few needles of newsworthy information in a PDF haystack is dizzying. Blearily tabbing through the files at random this morning, I came across screenshots of what appear to be emails between prosecutors in Epstein’s 2008 sex-crimes case, which resulted in Epstein getting a cushy plea deal (almost all of the names in the email are redacted). I can think of no reason that the names of those who afforded him such an arrangement shouldn’t be made public. In one of the emails, from late May of that year, one person mentions an unnamed person, presumably Epstein, spending only 90 days in jail. “Please tell me you are joking,” the other replies. “Maybe we should throw him a party and tell him we are sorry to have bothered him.” Such emails, although redaction-heavy, are the kind of information that journalists and investigators have longed for—they shed partial light on the government’s leniency in the case. Still, the release is piecemeal and difficult to comb through; as a result, it paints an unclear picture.

As is often the case online, the messy, public release has at times led to more confusion than clarity. On X this morning, I came across a viral post containing a screenshot of one of the FBI tips from the files that alleges that Trump and Epstein raped a woman. “Now we’re starting to see why Trump was hiding the Epstein files, and it probably gets much worse,” the post reads. Digging through the files, I’ve confirmed that the document is real, but the post—which currently has several million views—lacks crucial context. The allegations are not part of a court document or witness testimony; they’re transcribed from a 2020 call to the FBI tip line, and totally unconfirmed.

This is a sterling example of the informational chaos here. A disturbing, salacious tip, the credibility of which is completely unknown, printed on an official FBI form: It’s perfect fodder for screenshots, reposts, and accusations. The “information” looks terrible for Trump, but it’s presented without any burden of proof. That the DOJ would release something so potentially incendiary but redact other information, such as the names of government lawyers, only adds to the confusion.

The Epstein scandal has consumed Washington and dogged Trump’s second term, and the release of these latest files is textbook news dump: a massive tranche of individual image files and PDFs, collected with few discernible organizing details, dropped online just before the Christmas holiday. Deputy Attorney General Todd Blanche said on Sunday that the partial, phased release is being done in part to protect victims. Although that could be the case, the drip-drop release has the added effect of being frustrating and overwhelming, stringing everyone along during a moment when fewer people are likely to be paying attention.

Ironically, the nature of the release also means that the story will not die. Today’s release has only fanned the flames of the conspiracy. It also fragments an important news story such that it becomes hard to get a good sense of where things stand. In one interpretation, it might feel like the walls are closing in for Trump and the White House, where an avalanche of anecdotal evidence—the infamous 50th-birthday-book release in September, a trove of emails in November that mention Trump and his onetime adviser Steve Bannon, last Friday’s release, today’s—is piling up. But seen another way, this release is also optimally confusing, muddying the waters with as-yet-unverified information that’s being disseminated via individual screenshots on social media, making the whole thing easier to dismiss.

[Read: You really need to see Epstein’s birthday book for yourself]

There’s a secondary effect for those of us watching, which is that of being trapped in some kind of Epstein holiday purgatory. Family gatherings and Honey Baked Hams are colliding with the slow-burn proliferation of crime-scene evidence related to an alleged prolific sex trafficker who appears to have been close friends with the current president of the United States. Those following the Epstein saga closely are stuck waiting for the next shoe to drop; those with more normal news-consumption habits or who may wish to ignore the sordid affair may be forced to acknowledge it as nauseating details barge into their life while they scroll, channel surf, or talk with a politics-obsessed uncle at the dinner table.

This would be a small price to pay, if any true accountability were to come from this process. But much of the context of the Epstein files is that they are being released by a DOJ that, as my colleague David A. Graham wrote yesterday, has gone to great lengths to politicize itself in the second Trump administration. As Graham notes, the entire Epstein ordeal is a showcase of “compounding failures” by the federal government, from its slowness to act on tips about Epstein many years ago, to the plea deal in 2008, to this administration’s questioning of Epstein’s associate Ghislaine Maxwell and her move to a minimum-security prison this past summer. And then there is what the files continue to confirm: a moral rot in some of the wealthiest and most powerful people in the world. Combine these things, and the files are a recipe for inspiring potent distrust and resentment.

Those of us paying attention are, for now, stuck—bombarded with enough troubling information and allegations to assume the worst about this conspiracy, but also possessing enough earned cynicism and suspicion to assume that little will change.