Kaitlyn Tiffany | The Atlantic

The Great Travel Meltdown of 2026

2026-04-10T07:30:00-04:00

The Great Travel Meltdown of 2026 started taking shape at the end of February. At first, the U.S. war against Iran forced the cancellation or rerouting of many flights to the Middle East; then the blockage of the Strait of Hormuz drove up the price of jet fuel and threatened to cause crises for the major airlines. Though the two-week cease-fire announced last night may reopen the strait, prices are unlikely to rebound immediately.

Separately, large numbers of TSA workers started staying home after a protracted budget fight in Congress left them working without pay for weeks on end. Airport-security lines snaked into terminal basements or out their front doors. President Trump deployed ICE agents at the nation’s major airports, and although TSA workers are now receiving back pay, the funding situation isn’t yet resolved.

Getting somewhere by plane has always been an onerous proposition. If you search the phrase travel chaos on Google News, you will find that headlines about “travel chaos” reoccur in batches about every six months, going back to the beginning of time. But as a result of recent, tragic world events, the state of consumer aviation seems to be deteriorating at a rapid pace. Now Americans with travel plans would like to know exactly how worried they should be, and exactly how worried everyone else already is.

I’m one of the worriers. I’ve been planning to go to Barcelona for my honeymoon this summer. I’ve already read two books about the Spanish Civil War and just started a pretty dry one about the finances of the city’s famous football team. Last week I watched my fiancé spend every Capital One point in his account on our basic-economy flights, because the Google Flights trend line showed the fare for our trip going up, up, up, and headed off the chart.

[Read: The golden age of flying wasn’t all that golden]

So I’ve been in the forums—mostly on Reddit. People there are fretting about the known problems as well as interesting new ones that they came up with themselves. They’re worried, for instance, that an airline might decide to charge them an additional fuel fee upon arrival at the airport, and they don’t want to listen when someone replies, in an effort to be helpful, “Sounds illegal.” They’re worried about successfully flying to Japan but then getting stuck there by a fuel crisis that hits its peak with really, really bad timing (for them personally). In one thread, a commenter stated without explanation that “there is also a slim chance that events outside of our control will make people want to avoid air travel by this summer.” Okay!

Forum members rarely bother to acknowledge the insensitivity of stressing out over the effects of a distant war on your own summer vacation. But once in a while, someone’s post will push things just a little too far: It’s okay to worry that you won’t get to take a trip that you really care about, but it’s not okay to worry that if too many flights are canceled as a result of a distant war, you may lose your hard-earned gold status on the Australian airline Qantas.

Ominous reports of airlines’ crisis-management efforts have been attracting incredible attention. For many, the first big moment in this story was a March 20 memo from United Airlines CEO Scott Kirby that was sent to employees and then published on the company website—the type of thing an ordinary person would never read in ordinary times. According to the memo, jet-fuel prices had more than doubled since the start of the war. (Other sources have different numbers, showing that it had not quite doubled at that time.) Kirby presented this as a major challenge for the company—United might end up spending an extra $11 billion annually on fuel—but also, somehow, as a manageable one. “Demand remains the strongest we’ve ever seen,” Kirby wrote. He added that he was typing his note while listening to his son cheer during a college-basketball game, which he found inspiring. “There’s a part of me that can’t help but feel United is playing offense right now with potentially big rewards at the end.”

Maybe for an airline CEO, higher prices are their own reward. The travel experts I spoke with for this story said that summer flights will be really expensive. Airlines used to hedge against spikes in jet-fuel prices with preemptive financial maneuvers, but they don’t do this so much anymore. Now when fuel prices go up, they just raise fares for passengers instead. Some airlines have added fuel surcharges to the cost of each ticket (though this will be assessed at booking, not when you get to the airport). United Airlines is among those carriers that have raised the fees for checked bags, presumably to make up for some of its increased costs. Alli Allen, a travel adviser, told me via email that prices seemed to be escalating “by the minute!” Recently, she looked at flights for a client, found the price to be too high, and checked back 30 minutes later in the hope that maybe it had dropped. Instead she found that it had gone up by $300.

[Read: Flying is weird right now]

Clint Henderson, a writer and an editor for the popular website The Points Guy, said the same. “I think it’s going to cost a lot more for most people to travel this summer,” he told me. “Whether you’re using points and miles or cash, they’re all going to be higher.” He also expected the travel experience to be stressful, especially if TSA workers end up missing any more paychecks. Although news outlets, airlines, and the TSA itself (through the MyTSA app) offer tools to track security wait times, they can still be difficult to predict. Henderson said that he’d gone to check out the Atlanta airport at the height of the TSA-payment crisis and saw travelers facing an hour-and-a-half wait; then he went back the next day, and it was five minutes. “If this goes on, obviously it would be a disaster for the summer travel season.” When I asked him to rate the potential for chaos on a 10-point scale, he said he would give it a nine. (Take it from a points guy!)

Henderson said The Points Guy website’s official recommendation is that people book all travel for the year right now, even if it seems expensive, because conditions may only worsen over time. To avoid long lines, he also suggested flying out of smaller airports on Tuesday, Wednesday, or Sunday. The other travel trips that I accrued from emailing travel agents and industry bloggers will not impress you. They said to try to sign up for TSA PreCheck or apply for Global Entry, to show up at the airport early, and to bring snacks with you.

Travelers may be complaining, fretting, and catastrophizing, but so far, at least, they are doggedly proceeding with their plans. Airlines report that people are paying the higher ticket prices, and that the industry is seeing record levels of revenue. If Americans can go to Europe this summer, they will go to Europe this summer. And Europe (plus people from many other places) will come here. More than 1 million international travelers are expected to attend the World Cup. Matches will be held in several of the cities that have had the longest security lines, including Houston and Atlanta, and the final will be hosted in the New York–New Jersey area, which is home to the worst airport in America.

A new, more aggressive and pervasive form of travel chaos may yet ensue. In the meantime, though, behaviors are unchanged. Despite the rising prices, the spectacular security lines, and all of the rumored airport inconveniences, “we’ve seen very little evidence that people are canceling or toning down their summer travel plans,” Henderson said. “I’m constantly shocked by Americans’ insatiable demand for travel.”

A Writer of ‘Hypnotic Repetition’

2026-04-09T07:30:00-04:00

A love-it-or-leave-it feature of Nancy Lemann’s distinctive, dreamy style is that she often repeats herself. Images, events, and turns of phrase reoccur both within and across her five novels and even, to a lesser extent, in her nonfiction. The people in her books are always “falling apart”; their hearts are often “in a million pieces on the floor.” Her narrators—usually women from New Orleans—have a reverence for older traditions, including baseball, which represents “a chance to go forth with the heroes,” and for men in seersucker suits who have outdated affectations, like reading ancient Greek and eating oysters at lunch. Several of these women have soft spots for the same “blue-eyed boy with the crooked smile” (who has a drinking problem), and they tend to indulge in what could be called negative self-talk (they chide themselves for being “idiotic” or “lamebrained”).

In the introduction to a new reissue of Lemann’s 1985 cult-classic debut novel, Lives of the Saints, the British writer Geoff Dyer makes note of the “hypnotic repetition” that gives the book its rhythm and, crucially, its sense of place; he points to its frequent depiction of rain that slashes and nights that swelter. The repetition is also what, over time, has given Lemann’s work its poignancy, so it’s fitting that New York Review Books, which published the reissue, has simultaneously released Lemann’s first new novel in 24 years. The Oyster Diaries returns readers to New Orleans and revisits some of the stories from Saints, and it is partly about the humiliating experience of not seeing what’s right in front of us, or not understanding it, and having to look again and again. It has an air of running the tape back once more—this time, with the perspective gained by the passage of time and contact with a new generation.

The middle-aged narrator, Delery Anhalt, who bears many slant resemblances to Lemann, bounces between Washington, D.C., and New Orleans, and promises at the outset to recount “a bonanza of heartache that ultimately calibrated my soul with insight my soul had been waiting for its whole life.” This is a much more direct statement of purpose than Lemann’s previous narrators have given. The new novel also has a more concrete plot, though calling it plot-driven would be going too far. And it exists more firmly not only in its place but in its time, which is roughly right now.

When I spoke with Lemann in March, she had just returned from Argentina; she apologized for having “Argentine COVID,” which she worried was clouding her thoughts. Finding herself back in print, she seemed both grateful and a little disoriented, and she referred to the years that passed between the publication of her 2002 novel, Malaise, and these NYRB books—plus another publisher’s reissue of her 1987 nonfiction book, The Ritz of the Bayou—as “the Doom.” She regarded the process that brought the Doom to an end as somewhat magical. In fact, it was largely the doing of a publicist, Kaitlin Phillips, who started recreationally championing Lemann’s work several years ago. Phillips connected Lemann with an editor at The Paris Review and one at Harper’s, both of whom published new work of hers (which helped her get a new agent).

Lemann was born in New Orleans, and lived there until she attended college at Brown. “Before I left New Orleans, I just thought, Okay, this is what it’s like. This is what life is like,” she said. “But the minute I left and went up north I just realized, Oh my God, I have this ace in my back pocket.” The city was so singular and strange, its style so well preserved by its “isolation and remoteness,” that she could write about it endlessly. Lemann was 28 when she published Lives of the Saints, the story of an eccentric New Orleans family and a young woman experiencing her “wastrel youth.” Now she is 70. One of The Oyster Diaries’ two epigraphs is from the Greek poet C. P. Cavafy: “You won’t find a new country, won’t find another shore. This city will always pursue you.”

[Read: Twenty years after the storm]

Even when Lemann is not writing about New Orleans, she’s writing about how wherever she is resembles or doesn’t resemble New Orleans. New York’s “suave crumbling gleam” reminds her of New Orleans; Washington, D.C., where “everyone acts like federal tax bureaucrats, just by osmosis,” does not seem to remind her of New Orleans. In Malaise, the only Lemann novel narrated by someone not from New Orleans, the lead character—who is from Alabama—nonetheless compares Los Angeles’s Sunset Boulevard to New Orleans’s French Quarter, finding them both “darkly beguiling.” “People say I think everything reminds me of New Orleans, and it is true,” Lemann told me.

Mario Ruiz / Getty

Author Nancy Lemann returns to fiction—and her native city—with a new novel, The Oyster Diaries.

Across her work, the city has been embodied by a recurring character named Claude Collier, who was the love interest in Lives of the Saints. A modern reader might be tempted to roll their eyes at him; ungenerously, he is a failson. An alcoholic, he eventually drives his father’s law firm into the ground. He would fly through a casting call for the Bravo show Southern Charm. But Louise, Saints’ narrator, finds him angelic despite his vices, and continually points out that he is kind, wise, polite, tender, and lovely. (“He had the sweetness of the town itself and broke my heart completely into a million pieces on the floor,” she says.) Revisiting the novel now, Lemann found it a bit ridiculous that Louise was “constantly rhapsodizing” about this guy. “I can’t even get through that book when I try to reread it,” she said.

And yet she is clearly still fond of the rogue. She has continued to bring him back in her novels, even after saying in a 1988 interview that she would not. Claude was an out-of-frame instigator of the drama in her 1992 novel, Sportsman’s Paradise, and we learn the end of his story in The Oyster Diaries, after he reappears in the last third of the book (on a motorcycle). This time, he’s different, or he’s shown to us differently. Claude is still kind, wise, polite, tender, and lovely, but he’s now a middle-aged man whose life has caught up to him. “The new narrator admires him the way Louise did, but she’s not all sappy about it all the time,” Lemann said. “And she doesn’t have the stars in her eyes, you know?”

Though The Oyster Diaries finds space to close the loop on Claude Collier, it spends more time with August Anhalt, the narrator’s father and the keeper of the namesake oyster diaries. These are notations about the oysters he eats most days—“awful,” “still awful,” “terrible,” “no good,” “wonderful,” “exceptional,” or “beautiful to look at but no salt.” The plot takes off when the great tragedy of August’s life—a spousal betrayal—reoccurs by befalling his daughter. Unlike Lemann’s best-known heroine, Delery does not find everyone amazing; she finds most people annoying and moronic instead. But not her father: He is “rickety but suave—like New Orleans.”

Lemann knows that her debut novel, a romantic portrait of the deep South in the late ’70s and early ’80s, may now read as old-fashioned in certain ways. Lives of the Saints barely remarks, for instance, on the Black house staff who attend to the central characters. Lemann remembered a New York “beau” telling her to take these figures out, but she couldn’t do it without making the story unrealistic. Yet in The Oyster Diaries, Lemann widens reality’s lens a little. She directly describes the residential segregation of her beloved city. Her narrator, in the course of volunteer work as a court observer, records racial disparities she notices in the legal process; she also owns up to her own lifelong reluctance to look at such things directly, referring a couple of times to “the naked bulb” being too harsh.

[Read: ‘Intensely southern and only faintly Jewish’]

The Oyster Diaries isn’t a midlife-crisis novel, Lemann told me. It’s about the differences between youth and age—about death, on the one hand, and disillusionment, on the other. It’s about, as she put it, “having the veil rent from the temple, having the stars dashed from your eyes.” These revelations are coming pretty late in Delery’s life, Lemann admitted. “What world was she living in? Some childlike world.” Now her father is dying, her husband deceives her, and she lives in D.C. during an intense political era. Time to wake up.

Lemann has written about President Trump (“the world’s most obnoxious person”) before, and in this book he is something like an ominous general presence—a “Rasputin-like menace,” as she put it to me. Before Trump, she thought that politics were mostly boring, and that they were supposed to be. Now she works the voices of her adult daughters into her writing, often demonstrating the distance between the generations—a generation that could call politics “boring” without being reprimanded about privilege, versus one that can’t.

I found these moments of intergenerational friction to be some of the funniest of the new book. Lemann respects her daughters (one of whom, Emmeline Clein, is also a writer), but she also teases them. She sketches some great scenes in which the younger generation’s sincere sense of moral responsibility and obvious correctness on the facts clashes with the older generation’s often-overlooked strong suits. The latter include a creatively productive ambivalence, an ingrained reluctance to make other people feel stupid, and the ability to just talk about something else at dinner. “They try to teach me and I learn and I accept their teachings,” she said of her daughters. “But I also like to satirize their teachings sometimes.”

She satirizes herself, too. A late stretch of The Oyster Diaries is set on a family trip to several countries in Africa, and begins with Delery noting, “I always knew Africa would remind me of New Orleans.” New Orleans again?, you have to ask. But that seems to be the joke. “My daughters would say I should explore, study, and unlearn certain aspects of these sentiments,” the narrator adds fondly. Maybe she will; maybe she won’t. Maybe Lemann will write another odd, wonderful novel, and we’ll find out.

A Legal Decision That Could Change Social Media

2026-03-25T15:40:22-04:00

Updated at 8:48 p.m. ET on March 25, 2026

After deliberating for nine days—and emerging at one point to tell the judge that it was having a difficult time reaching a decision—a jury in Los Angeles finally returned its verdict today, finding both Meta and Google liable for creating addictive products that caused a young woman’s mental-health problems.

The two companies were ordered to pay $3 million in compensatory damages: 70 percent by Meta and 30 percent by Google. (Meta-owned Instagram played a larger role in the complaint than Google-owned YouTube, which explains the split.) This is hardly any money to either of these companies—Meta alone brought in nearly $60 billion in revenue over the last three months of 2025. But the verdict will lead others to pursue similar cases against tech companies (thousands are already pending), and possibly result in changes to the design of social-media apps.

Following the verdict’s announcement, Matthew Bergman, one of the plaintiff’s lawyers and the founding attorney of the Social Media Victims Law Center, sent a lengthy statement to reporters. “This verdict carries implications far beyond this courtroom,” it read in part. “It establishes a framework for how similar cases across the country will be evaluated and demonstrates that juries are willing to hold technology companies accountable when the evidence shows foreseeable harm.”

A Meta spokesperson sent a shorter statement just after the verdict was read: “We respectfully disagree with the verdict and are evaluating our legal options.” In a later email, the company updated its statement, saying it would appeal the verdict. It also said: "Teen mental health is profoundly complex and cannot be linked to a single app." Google will also appeal, according to the spokesperson José Castañeda. “This case misunderstands YouTube, which is a responsibly built streaming platform, not a social media site,” he wrote in an email.

The plaintiff in this case, a 20-year-old named Kaley, was referred to in case documents by her initials, KGM, because the events she was suing over happened when she was a minor. She originally filed against TikTok and Snap as well but settled with them before the trial.

The core questions of the case were whether the social-media platforms had been designed to be addictive, and whether a social-media addiction could be said to have played a direct role in causing the mental-health issues that KGM experienced as a child. In her complaint, she said she had a “dangerous dependency” on the platforms and that they had contributed to her “anxiety, depression, self-harm, and body dysmorphia.”

Today’s news comes right on the heels of a verdict against Meta in another case, brought by the New Mexico Attorney General Raúl Torrez, which was announced yesterday. The jury for that trial agreed that Meta should pay a penalty of $375 million for thousands of violations of the state’s consumer-protection laws. The issue at stake there was relatively specific: The state argued that certain design and moderation choices left kids vulnerable to online predators on Meta platforms and indirectly enabled serious crimes. The facts were highly technical and, unlike the Los Angeles case, didn’t involve qualitative assessments of young people’s personal lives or thorny debates about whether social media can be addictive.

Yet, it was a telling verdict and a massive judgment. Torrez emphasized its significance in a statement to reporters, writing, “New Mexico is proud to be the first state to hold Meta accountable in court for misleading parents, enabling child exploitation, and harming kids.” Meta plans to appeal the verdict, and sent its own statement to reporters yesterday, which read in part: “We work hard to keep people safe on our platforms and are clear about the challenges of identifying and removing bad actors or harmful content. We will continue to defend ourselves vigorously, and we remain confident in our record of protecting teens online.”

[Read: How Meta executives talked about child safety behind the scenes]

KGM’s case was novel because it treated YouTube and Instagram as fundamentally defective products. The issue wasn’t whether bad actors could exploit them but whether the platforms themselves were dangerous. Online platforms are generally not legally responsible for the content that their users post; Meta, for example, would not be liable for bullying comments or imagery for self-harm posted onto Facebook. But the judge in this case, Carolyn Kuhl, decided that design features such as algorithmic feeds, auto-playing videos, and push notifications were valid targets. Members of KGM’s legal team successfully argued that Instagram and YouTube were created by companies that knew they were addictive and harmful and that chose not to warn consumers.

Though most people would usually think of product liability as applying to things such as poisoned baby powder and cars without seat belts, the idea here is that social media can have effects as tangible as those of physical goods, and we should think about it in the same terms. Such metaphors abounded in the trial. Mark Lanier, a member of KGM’s legal team, described social-media companies as lions hunting gazelles, and compared their products to cigarettes, the free tortilla chips that patrons may mindlessly snack on at a restaurant, and the baking soda in a cupcake. The baking-soda metaphor was meant to underscore that Instagram and YouTube had an outsize effect on KGM’s life, the way a tiny teaspoon of baking soda competes with more substantial ingredients such as flour or eggs in a cupcake recipe. But it was KGM’s own account of her experiences that appeared to move members of the jury, some of whom reportedly cried during her testimony.

Coverage of the case had died down significantly after the newsy high point of Mark Zuckerberg’s testimony in mid-February, but a handful of reporters provided updates from Los Angeles. Both sides found expert witnesses who offered competing accounts of whether social media can literally be said to be “addictive.” The lawyers also told competing stories about what caused this one girl’s mental-health problems.

Whereas Google’s closing arguments focused on whether KGM was actually addicted to YouTube and whether YouTube is more similar to television than it is to social media, Meta’s lawyers emphasized the other problems in KGM’s young life, including her fraught relationship with her mother and her older sister’s hospitalization for an eating disorder. They also called to the stand her former therapists, one of whom said that social media had rarely come up in their conversations. Another said that she believed that social media was “a contributing factor” in KGM’s anxiety, though not its primary cause. In his closing argument, Meta’s lawyer Paul Schmidt insisted that KGM’s representation had needed to prove that taking Instagram out of her life would have made it “meaningfully different.” They didn’t do that, he said, though the jury apparently believed otherwise.

The case in Los Angeles was only the first of many—“a brick in a potential wall,” as the Cornell law professor James Grimmelmann put it when the trial began. In fact, Meta and other social-media companies are facing so much pending litigation that keeping track of it all can be hard. Jury deliberations in Los Angeles were simultaneous with those in New Mexico. The company will be a defendant in another upcoming bellwether trial in the Los Angeles court, this one filed on behalf of a minor identified by the initials RKC, who similarly claims that he became addicted to social media and that it caused him to experience suicidal ideation, body dysmorphia, anxiety, and depression, “among other harmful effects.” That trial is expected to start this summer. And at the same time, an enormous multi-district litigation incorporating thousands of personal-injury suits against major tech companies will proceed in Oakland, starting with a Kentucky school district’s complaint that social media has been so poorly age-gated and so distracting to young students that it has effectively become a public nuisance.

[Read: No one knows exactly what social media is doing to teens]

In these upcoming cases, new juries will be considering entirely new sets of personal facts, but they’ll also be considering the same basic questions about addiction, liability, and cause and effect. Of course, future juries may understand those issues differently than those who reported back this week. These questions are complicated, which is why we’ve ended up in the strange situation of hearing them argued in court rooms in the first place. Many have compared this succession of lawsuits to those that took down Big Tobacco in the 1990s, though experts have also pointed out that the comparison between social media and cigarettes is not very exact. (“We’re not talking about a biological substance that you can consume that has a demonstrable chemical effect,” Pete Etchells, a professor of psychology and science communication at Bath Spa University, in England, told me in January.)

Still, social media has clearly reached a fork in the road. The existential questions that all of these lawsuits are asking are whether it is possible for social-media platforms to directly cause mental-health issues and other serious, life-changing problems for young people, and whether it is feasible to hold the companies behind them accountable for that. The upcoming trials likely will not bring us to a totally satisfying answer on the first, but they will certainly shed a lot of light on the second.

What Was Grammarly Thinking?

2026-03-12T12:45:48-04:00

To me, the best first sentence of any piece of journalism is the one in Joan Didion’s 1987 book, Miami, which begins like this: “Havana vanities come to dust in Miami.”

I love that sentence and that propulsive first chapter so much that I once sat down to try to figure out how she did it. I looked at the sentences one at a time to assess what purpose each one was serving, and I counted how many of them Didion had needed to accomplish each thing she wanted to accomplish. Then I thought about how she figured out what order to put them in to have maximum page-turning impact. And then I compared all of it unfavorably with the flailing and feeble way in which I would have pursued the same goals. I marked up my copy of the book in a somewhat desperate fashion and then became depressed.

That type of copying is pretty normal, and they teach it in school. It’s how you learn (and how you become depressed). But in the age of generative AI, there are many new kinds of copying. For instance, Wired reported last week on a tool offered by Grammarly, which briefly offered users the opportunity to put their writing through something called “Expert Review.” This produced AI-generated advice purportedly from the perspective of a bunch of famous authors, a bunch of less-famous working journalists (including myself, per The Verge’s reporting), and a bunch of academics (including some who had recently died).

[Margaret Atwood: Murdered by my replica?]

I say “briefly” because the company deactivated the feature today. A lot of people got really mad about it because none of the experts had agreed for their work to be used in such a way, or to serve as uncompensated marketing for an app that people use to help them write more legible emails. “We hear the feedback and recognize we fell short on this,” the company’s CEO, Shishir Mehrotra, wrote on his LinkedIn page yesterday. Not long after, Wired reported that one of the journalists whose name had been used in the feature, Julia Angwin, was filing a class-action lawsuit against Grammarly’s owner, Superhuman Platform. In a statement forwarded by a spokesperson, Mehrotra repeated apologies made in his LinkedIn post and added, "We have reviewed the lawsuit, and we believe the legal claims are without merit and will strongly defend against them.”

Before the tool went down, I spent a few hours experimenting with it, trying to see what it might be like to be edited by myself. I was hesitant to do this, because I had once asked ChatGPT to write something as if it were me (just for fun!) and found the experience humiliating. The result was sentimental and ditzy—it was studded with cloying rhetorical questions, had a bizarre number of unnecessary exclamation points, and sounded exactly like me.

But I still wondered, out of self-obsession, how an AI imitation of me might advise the real me if I fed it prose that I had written, and whether it could possibly make that prose better. Clearly, this experiment was sort of a gimmick. I assumed the suggestions would exist on a spectrum from obvious to dumb, though I was open to being surprised. If I’m being honest, what I was most interested in was seeing who I am in this latest iteration of The Computer. I also wanted to see whether the tool was good enough that someone might someday use it instead of hiring a human editor. If it was, I would have to have a difficult but compassionate conversation with my boss.

To my dismay, I was unable to summon the AI version of myself. I pasted in numerous articles I’d written and numerous fake articles that I had asked a chatbot to make up. But Grammarly seemed to think other writers were more expert in these articles’ subject matter and therefore more qualified to advise me. It suggested tech journalists, pop-culture academics, and legendary practitioners of narrative nonfiction. I wouldn’t appear. My boss tried too. He messaged me: “i have both claude and chatgpt writing fake essays in an attempt to fool a different AI into presenting me with an unauthorized simulacrum of one of my writers.” He failed. We both felt bad about the way we were spending our time.

So I gave up on that and started engaging with the experts I had been given. The tool was really pretty funny. It was not impersonating people in exactly the way that I’d imagined it would. I wasn’t getting a message from a bot pretending to be the New Yorker writer Susan Orlean. At no point did Grammarly say, “Hi, I’m Susan Orlean.” Instead, it would say, “Taking inspiration from Susan Orlean,” “Applying ideas from John McPhee,” “Using concepts from Bruce V. Lewenstein” (an undergraduate professor of mine, coincidentally), and so on.

[Stephen King: My books were used to train AI]

The inspiration, ideas, and concepts that the tool drew from these writers and thinkers were, with no exception, incredibly stupid and unhelpful (thank God). When I pasted in a story that I had written about TikTok, for instance, Grammarly told me it was drawing inspiration from my co-worker Charlie Warzel’s Galaxy Brain newsletter and then suggested changing the headline from “TikTok’s New Paranoia Problem” to “TikTok’s Zeroed-Out Voices: The New Paranoia Problem.”

When I asked it to look at an excerpt from my 2022 book on One Direction fans, it told me that it was going to improve the first sentence with a suggestion inspired by Joan Didion’s The White Album. Amazing! But then the idea was just to open with a quote from a young woman I had written about, which didn’t seem uniquely Didion-esque. The bot clarified. “In The White Album, Joan Didion emphasizes the importance of personal narratives in understanding reality, stating, ‘We tell ourselves stories in order to live.’” (As you may know, this super famous and often-misquoted line actually refers to how we have to delude ourselves constantly in order to stave off the certainty that all is meaningless.) Then it made up a fake quote that I might consider using.

I was sometimes offered suggestions inspired by the sociologist Sherry Turkle or by the famed memoirist Mary Karr. But for some reason, Grammarly offered suggestions inspired by the essayist Leslie Jamison over and over, almost insistently. I heard from both “Gia Tolentino” and the New Yorker writer Jia Tolentino. None of the suggestions was about structure, organization, or trimming the fat from a story. All of the suggestions were wordy additions. Some were needlessly floral elaborations and fabricated details clearly meant to add color and voice. For instance, a long and fake story about my late grandmother appeared in the middle of one draft. Others were stilted explainer-y tangents that seemed written for readers with no preexisting knowledge of the world. One idea was to pop a several-sentence capsule history of the entire feminist movement into the middle of a paragraph that mentioned the “girlboss” trope inspired by the philosopher Amia Srinivasan.

I tried to talk with the chatbot integrated into Grammarly about the situation, but it had no idea what I was asking about. It insisted that Expert Review was done by anonymous human editors, none of whom was famous, and assured me that Grammarly would never claim to be Joan Didion while giving me advice. We had a confusing exchange about that for a while before it revealed that its knowledge of the world and its own platform went up only to June 2024. Soon after, I learned that someone else had asked the tool to do an Expert Review on a bunch of “lorem ipsum” nonsense text and that it had obliged with recommendations inspired by Stephen King. (And then, as mentioned, the CEO killed it via LinkedIn.)

Now that I’ve looked more closely at this not-very-useful feature, and now that it’s shut down, the whole situation seems a little absurd. This was just a weird and inappropriate thing that a company tried to do to make money without putting in very much effort. The primary reason it became a news story at all was that it touched on widespread anxiety about whose work is worth what, whose skills will continue to be marketable in the age of AI, and whether any of us are really as complex, singular, and impossible-to-imitate as we might hope we are.

When I started working in journalism, in 2015, commenters (usually men) would reply to my stories and tell me to “learn to code.” This was a common taunt and catchphrase of the era (Gamergate), and it was a nod to the massive cultural, political, and economic shifts under way at that time. Tech was ascendant in every sphere, its hard skills were worth more money than ever before, and people like me—people who knew only words—seemed soft and useless in such a world.

Lately, there have been rumblings about a reversal. Large language models are very good at things such as coding, programming, and dealing with numbers. Users on X recently resurfaced a 2024 interview clip in which one of the most influential technologists of our time, Peter Thiel, said he thought the post-AI labor market would actually be “much worse for the math people than the word people.”

You might think I’m bringing that up to boast about how I came out on top in the end—it all worked out for me, and the latest AI failure proves that no bot can do what I do and no bot ever will. That’s not what I’m saying. What I’m saying is that the “learn to code” guys committed the crime of hubris, but I won’t.

A Never-Ending Conspiracy Theory in Remote Alaska

2026-03-10T07:00:00-04:00

The guy pouring my beer in Anchorage told me that he knew there was no truth to decades-old rumors about a research facility 200 miles to the northeast. Nobody was up there talking to aliens or controlling people’s minds. “They just do the aurora,” he said, cheerfully, while tearing up pieces of mint.

The comment didn’t surprise me. Many people who don’t believe one conspiracy theory about that station—known as the High-frequency Active Auroral Research Program, or HAARP—believe another. A common misconception is that it can manufacture northern lights, a natural wonder typically most visible in or near the Arctic Circle. It cannot (and neither can any man-made instrument). Still, late last year, when a geomagnetic storm caused aurora sightings as far south as Texas, Facebook was studded with posts warning that these lights were not “natural” and that they were created by the scientists at HAARP for possibly sinister reasons.

I’ve been curious about HAARP for a while because of rumors such as this one. The lab has also been erroneously credited with various supernatural occurrences (backward-walking caribou) and secret contact with extraterrestrials (covered up by “men in black”). Most commonly, it’s blamed for events caused by nature. The office phone rings after hurricanes, earthquakes, floods, wildfires, tornadoes, and typhoons, no matter where in the world they occur. A 2024 study found that HAARP was the subject of more than a million conspiracism-inflected posts on Twitter from January 2022 to March 2023, primarily about natural disasters. In early 2024, the far-right influencer Laura Loomer suggested that HAARP created a snowstorm to dampen turnout at the Iowa caucuses and thwart the Trump campaign. And when I visited HAARP this past November, calls were coming in about whether the facility had caused Hurricane Melissa, which had recently swept through Jamaica and Cuba, resulting in at least 88 fatalities and billions of dollars in damage.

[Adam Serwer: Gullible, cynical America]

All of this anxiety is focused on a unique research instrument housed at HAARP, which is owned by the University of Alaska at Fairbanks and was originally built by the military for the cost of $290 million. “The array,” as the instrument is called, is a grid of 180 transmitters that each sit atop a 72-foot-tall post, arranged in a clearing and surrounded by Alaskan wilderness. You could call it the world’s highest-powered radio transmitter, but it’s more precisely its most powerful ionospheric heater (which sounds scarier). HAARP transmissions reflect off of the ionosphere—part of the upper atmosphere that starts about 30 miles above the Earth’s surface—and temporarily “heat” or excite it.

The Navy hoped to use the facility to work out new forms of long-range communication, and the Air Force wanted to study “killer” electrons that sometimes damage satellites. But their interests in these pursuits ran out, and the military turned the facility over to the university in 2015, rather than bulldoze it. David Hysell, an engineering professor at Cornell who has conducted experiments there, told me that the most succinct way to summarize what HAARP now studies is “the effects that the ionosphere has on signals, on radio-wave propagation,” which is not very exciting. The equipment looks crazy, but it can’t affect the parts of the atmosphere where the Earth’s weather is created.

Still, the calls to the lab continue. The Facebook posts go viral. The university has held open houses, posted public information pages, and produced irreverent merch, but nothing seems to tamp down suspicion. Jessica Matthews, HAARP’s director, is an Air Force veteran, and her first instinct was to deal with conspiracy theories in the style of the military: “If left to myself, I wouldn’t say anything,” she told me. “But that’s not the right answer.”

She spent close to four hours on Zoom walking me through more than 100 slides about HAARP and its history. Afterward, I accepted her invitation to Alaska to see what it looks like for her team to contort around conspiracy theories every single day. How do memes and menacing phone calls shape the lives of a bunch of regular people doing confusing science in a remote town that has not even one good place to buy a pizza?

After flying into Anchorage, I drove for four hours through snowy mountains to a tiny bed-and-breakfast about 30 miles south of HAARP (the closest hotel had burned down the week before). Matthews and I caravanned to the facility at 6:30 the next morning, when it was still pitch-black outside.

The operations center—a windowless box in the deep, dark woods—was very X-Files on the approach. But inside, it was as boring as any workplace. There was a kitchen with a coffee maker and a refrigerator with mayonnaise. A square ring of offices surrounded a conference room; in the conference room, a small group of scientists joined a Zoom call with HAARP’s space-weather consultant, Whit Reeve. During the meeting, he informed the scientists about some minor geomagnetic storms and solar winds, unlikely to affect any of the day’s experiments.

A member of the maintenance team walked me through the HAARP power plant as he turned on its five generators, labeled Angel 1, Angel 2, Angel 3, Angel 4, and Angel 5. The current staff doesn’t know who named them, but it was clearly in reference to the 1995 book Angels Don’t Play This HAARP, co-authored by Nick Begich Jr., a member of a prominent political family in Alaska. (His son, Nick Begich III, is the state’s current representative in Congress.) Among other things, the book suggested that HAARP would “boil the upper atmosphere” with radiation that would then bounce back to Earth and “penetrate our bodies, the ground and the oceans,” and its publication instigated many of the early conspiracy theories.

[Read: The truth about hurricane geoengineering]

To be clear, I realize that if there were any reason to worry about what (currently) happens at HAARP, then scheduling a visit, walking through the front door, and being led on a friendly, guided tour would not be a very good way to assess those concerns. But I have found no reason to believe that the things the facility is accused of could possibly be going on. All of the experts I spoke with agreed that the charges were scientifically impossible. They also agreed that what really happens there is hard to explain.

I was at HAARP for the start of a weeklong “research campaign”—one of a few that the facility hosts each year. A modest grant from the National Science Foundation provides most of HAARP’s operating budget; for efficiency’s sake, the center groups experiments together and runs them back-to-back in these concentrated blocks. Such experiments mostly look like nothing—opaque data on a computer monitor—and result in papers with confusing titles about “whistler mode” waves and “ionospheric turbulence.”

Once in a while, the experiments look like something (and still result in papers with confusing titles). When the bartender in Anchorage told me that HAARP “does the aurora,” that was wrong. But it can make something like the aurora, in very small spots. The array can heat a patch of the ionosphere—or excite its electrons—to the point where it creates what scientists call “airglow.” This is an extremely faint, temporary aurora, of sorts, which can flicker red or green. You’d have to really stare at it, train your eyes, and convince yourself you were seeing it, in order to be impressed. More likely, you’d be underwhelmed. “HAARP is powerful by comparison to other things that humans do, but it’s nothing compared to what nature does,” Hysell, the Cornell professor, told me. The effect of a transmission is brief; everything goes back to normal within seconds.

But while I was there, there was no airglow. Just the invisible radio waves.

Matthews had prepared a chart for me of everyone present at the facility, color coded so that I would know who had agreed to speak to me on the record and who would rather their names not appear in an Atlantic story about the 30 years of conspiracy theories surrounding their workplace. When she and I sat down to talk in the conference room, it became clear that she feels a strong sense of personal responsibility for everything that happens around her. If anybody felt annoyed or uncomfortable about my being there, that would fall on her. If a magazine story produced a burst of negative attention that inconvenienced or endangered the people who worked for her, that would fall on her, too.

Leading up to research weeks, Matthews and the university’s communications team generally won’t post on social media or issue press releases. They’ve learned that attracting attention is not worth it. In 2023, Matthews approved a press release about an upcoming HAARP experiment that would create an airglow. A local became alarmed and showed up at the gate to the facility, where they filmed employees and then followed one for some distance down the access road. “Ever since then, I’ve been extremely gun-shy about putting any press releases out,” Matthews said. (The university has continued sending out “transmission notices,” or bulletins listing what times and at what frequencies the array will be transmitting, directly to citizen-science groups.)

[Read: Who’s ready to think about blocking out the sun?]

Whereas some of the calls that Matthews fields about HAARP are a bit funny—she often hears from people asking for nice weather on their wedding day—other calls are alarming. HAARP once received a bomb threat just before dozens of college students were set to show up for a weeklong program. (The FBI investigated and found that the suspect was not in the state.) And in October 2016, two men were arrested in Georgia with thousands of rounds of ammunition and numerous guns, including AR-15s, en route to HAARP to attack it. Those men got nowhere close to HAARP (and seemingly had no good plan to cross the border into Canada). Still, HAARP’s security adviser, Sean McGee, the former head of the university police, called it “a bit of an eye-opener.”

Employees are now trained in de-escalation strategies. McGee prepares staff for uncomfortable conversations (and possible confrontations) by encouraging them not to dismiss people’s concerns. “You’ve got to take what they say seriously,” he said. He told me about an interaction he had when he was a police officer. A woman was convinced that she was being spied on through a camera hidden in her smoke detector. McGee examined the device. “I assured her that there was no camera,” he said.

The woman then wondered whether there was a spy camera inside her TV remote instead. “That was okay,” he said. “I was able to take the batteries out of that.” The point of the story, he said, was that the kind thing to do is invest a little time and answer questions as best you can. Though I heard another takeaway: The questions might not ever end.

When HAARP runs during the day, the naked eye or ear has little way to know that the array is transmitting. You can even walk between the poles, thanks to a protective metal net that prevents the transmissions from bouncing to the ground. On a calm, frigid afternoon, Matthews showed me the array and pointed out that the aircraft radar was on, to make sure that no errant pilot tried to fly overhead. The instrument was, in my opinion, kind of beautiful, in the way that large, confusing objects sometimes are.

The questions that people have about HAARP generally fall into two broad categories, Matthews said. Outside of Alaska, in what Alaskans call “the lower 48,” people tend to be interested in conspiracy theories involving aliens, mind control, or large-scale disasters. Locals tend to have more personal questions. For example, a woman who attended a HAARP event asked whether the array had caused her neighbor’s cancer. “That’s a difficult thing to answer,” Matthews told me. The question prompted her to commission an independent study by a Seattle engineering firm, which showed that HAARP’s radiofrequency emissions were safe. Because of her military background, Matthews also feels obligated to respond directly to phone calls from veterans who seem to be in some kind of mental or physical distress, and to try to talk them through their concerns.

People who hear about HAARP today do so in an information environment that is extremely hospitable to paranoia. Climate change is causing an increase in extreme weather events, which provoke conspiracy theories consistently; certain pockets of the internet not only believe that the government controls the weather but now insist that it has replaced the sun with an LED lamp. The Trump administration stokes paranoia all the time, in innumerable ways. Over the summer, the president’s EPA administrator, Lee Zeldin, promised answers to “good faith” questions about the condensation trails behind planes. Currently, Republican lawmakers are calling for the passage of laws that would ban weather modification, placing all kinds of ordinary scientific experimentation under that scary-sounding heading.

On my second day at HAARP, the experiments were scheduled to run past dinnertime. Someone had brought a huge pot of chili and let it simmer on the stove in the shared kitchen, and everyone joked about that scene in The Office with the chili, and how important it had been to prevent the same catastrophe from happening here. While I spoke with Evans Callis, HAARP’s research-support-services lead, we were offered slices of a birthday cake that had just been presented to a member of the maintenance crew.

Callis had thought a lot about the conspiracy theories. He started volunteering at HAARP open houses as a college student and was hired by Matthews after he graduated. Around the time that he started the job, he was unsettled when anti-HAARP graffiti appeared on the side of his church, but it turned out to be a coincidence (he thinks). “We have this innate desire to want to control things, and there’s some things that just are beyond our control,” he told me. “We can’t control whether or not a typhoon forms in the Pacific.” We meaning a small group of Alaskans, and we meaning any of us.

When extreme weather causes mass destruction or death, he said, “it’s natural to ask, you know, Why?” If someone thinks the wrong people are controlling the weather and using their power to inflict misery, that may make them feel helpless and frustrated, he thought. But in a strange way, it may also make them feel less helpless and frustrated than if they imagined that there was no reason for anything.

With each passing year, HAARP is divorced a little further from its military past. Decades ago, the Air Force purchased the land surrounding the facility from the Ahtna people, under the threat of eminent domain, but it has recently sold it back. HAARP will soon be renamed the Subauroral Geophysical Observatory, or SAGO, matching the name of its NSF grant. “There is a side benefit that it helps us transition off of the name that drives conspiracy,” Matthews added.

Some people will likely point to the change as evidence of a cover-up. That’s probably unavoidable, given the facility’s setting. When I asked Whit Reeve, the space-weather consultant, why he thought that so many conspiracy theories about the array persisted, he guessed that the answer was obvious: “Well, one thing is it’s in the middle of nowhere and it’s got a fence around it,” he told me.

Throughout the long, true history of secretive military science, the government really did favor sparsely populated areas (and specifically Alaska) to avoid prying eyes, which has had the ironic knock-on effect of making any distant outpost seem like it’s up to something nefarious, even when nothing very interesting is going on. If HAARP were located just outside of Cleveland, maybe nobody would care about it.

Living with a conspiracy theory means thinking about this kind of thing all the time. Should the university share really cool photos of the array with aurora visible above it, or is that inviting a rash of Facebook misinformation? Should it sell merchandise featuring cartoon aliens, or is that making too much fun of people’s concerns? After dark, Matthews and I drove around the facility grounds, and she stopped in front of the array to point to the red lights at the base of each of the poles, which indicate that the transmitters are operational. They gave the entire scene an ominous glow.

“Would you post a photo of that on social media?” she asked. I considered it. It looked like something that could freak a lot of people out, I told her. It was kind of freaking me out just sitting there. No, I thought. I probably wouldn’t.

How Meta Executives Talked About Child Safety Behind the Scenes

2026-02-27T12:25:59-05:00

Updated at 2:48 p.m. ET on February 27, 2026

Not so long ago, Mark Zuckerberg was working in overdrive to convince the world that his company was doing everything it could to protect children. In 2021, he posted a note to his personal Facebook page, writing that he had “spent a lot of time reflecting” on the types of experiences he would want his daughters, then 4 and 5 years old, to have online. “It’s very important to me that everything we build is safe and good for kids,” he wrote, emphasizing that the company absolutely does not “prioritize profit over safety and well-being.”

But documents recently viewed by The Atlantic show that behind the scenes, the company now known as Meta was divided on whether protecting kids should take precedence over user growth and engagement. For years, the company only incrementally rolled out restrictive safety features, even as its own staff detailed the risks its platforms posed to children. Take, for example, a technical problem that affected the company’s systems in November 2020. This issue limited Meta’s ability to track bad actors, at a time when there were, according to an internal chat, “thousands of minors” reporting what the company refers to as tier-one “Inappropriate Interactions with Children,” or “IIC T1”—the “most severe” outcomes possible, such as meeting for sex in real life, suicide, extortion, sadism, and sex trafficking.

“Even though we know that there is IIC T1 going on (more than 50% of which is sextortion which can lead to suicide) we haven’t done anything. we had a broken escalation path and no measurements,” one employee wrote in the internal chat about the problem. “God knows what happened to those kids.” The company fixed the technical failure within weeks, another document shows, but it would take several more years to adopt other suggested measures to tackle broader issues that allowed predators to find underage targets on Instagram, which Meta owns.

Company spokespeople were clearly aware of the broader teen-safety problem. Just four weeks after Zuckerberg had posted about being “good for kids,” two public-affairs specialists discussed an Instagram update that had just rolled out. That update made new accounts belonging to 13-, 14-, and 15-year-olds “private” by default, yet even this modest move had been flagged by insiders as a business risk for nearly two years before the change was made. Liza Crenshaw messaged her colleague Sophie Vogel that the move had been “contentious”—Instagram CEO Adam Mosseri, a deputy to Zuckerberg, was concerned that it would cause a “huge growth hit,” Crenshaw wrote, according to documents we reviewed.

“We will never get out of this mess if he/we’re not just prepared to ERR ON THE SIDE OF SAFETY,” Vogel wrote. “Would he want any tom dick or harry being able to see all his kids’ content, follow them etc? Is he fucking nuts?”

Meta’s failure to proactively limit its business ambitions on behalf of vulnerable users may seem obvious to those who have paid attention to the company over the years, especially after revelations such as those leaked by whistleblower Frances Haugen in her “Facebook Files,” among others. Yet these new documents—which include corporate reports, presentations, and other communications and which were disclosed as part of a lawsuit against Meta that went to trial this month in New Mexico—provide an unusually clear view into the company’s decision making. Over the course of six years, Meta tinkered with basic privacy controls, while simultaneously calculating how simple interventions would moderately reduce the amount of time people spent on Instagram and initially opting for conservative, piecemeal updates to protect its engagement numbers.

[Adrienne LaFrance: ‘History will not judge us kindly]’

“It isn’t just that there is a neutral space that is created” on Instagram and Facebook, New Mexico Attorney General Raúl Torrez, who brought the lawsuit, told us in an interview last Thursday. “It is that existing design choices amplify the harm.”

As far back as 2019, employees had specifically run a test on Instagram to understand how experimental accounts engaging in what the company termed “groomer-esque behavior”—following “teen hashtags” or “sexy teen accounts”—would come across minors. At issue was the app’s recommendation algorithm, which connects soccer fans to pictures of Lionel Messi, for example, or aspiring travel influencers to videos about little-known cafés in Greece. It also seemed to funnel children to potentially dangerous adults with whom they wouldn’t otherwise be connected. “We are recommending nearly 4X as many minors to groomers (nearly 2 million minors in the last 3 months),” the report read, according to an internal document we viewed. According to this test, 27 percent of the recommendations shown to these “groomer-esque” accounts belonged to minors, compared with 7 percent of the accounts recommended to everyday adults. The report continues: “22% of those recommendations resulted in a follow request”—meaning that potential groomers attempted to interact with these minors nearly a quarter of the time. Even so, Instagram waited years before locking down accounts belonging to its youngest users.

Meta continues to dispute that it prioritizes its business over the safety of users, arguing that the documents show the company investigated and responded to safety threats as they emerged. While testifying in a different case in Los Angeles this month, Mosseri said, “We, I think over and over again, made changes to the platform that I think hurt revenue in the short term, but I think are not only good for people’s well-being but also good for the business over the long run.” And Andy Stone, a spokesperson for Meta, told us that the New Mexico lawsuit is baseless.

[Read: Can Instagram ruin your life? The jury will decide.]

“While the New Mexico Attorney General makes sensationalist, irrelevant, and distracting arguments by cherry picking select documents, we’re focused on demonstrating our longstanding commitment to supporting young people,” Stone said in a written statement. “For over a decade, we’ve listened to parents, worked with experts and law enforcement, and conducted in-depth research to understand the issues that matter most. We use these insights to make meaningful changes—like introducing Teen Accounts with built-in protections and providing parents with tools to manage their teens’ experiences.”

In August 2020—a year after the research on “groomer-esque” accounts but a year before Zuckerberg’s post about being a concerned father—Meta’s Growth Graph team created a slideshow to explore the question of whether teen Instagram accounts should be set to private by default. This would shield teens from unwanted attention by limiting the ability of people who do not know them to see their content or their profiles, or to contact them. As the Growth Graph team explained in the document, the move would “help prevent high severity actions such as child grooming and inappropriate contact with minors.” (Though the presentation referred to minors generally, Stone told us that at the time, Meta was particularly focused on users under the age of 16.)

The company’s legal, public-affairs, policy, and well-being teams all supported the change, as did teen users and their parents, the document asserted. “Parents are worried about the security and privacy of information and who can contact them/their teens,” the document stated. “Most teens prefer private accounts and wish to see privacy controls during onboarding.”

But internal tests showed that setting these accounts to private by default would lead to “serious growth and engagement decreases,” the document continued. Taking dramatic action to protect teens would mean fewer new teens signing up, existing teens using the platform less, and an overall drop in activity that the employees who created the presentation expected would compound over time. They presented an analysis that showed that overall time spent by teenagers would drop by 1.9 percent by the end of a five-year window. The growth team opposed the change, according to the presentation, which describes its position as “Don’t Launch (Now).”

This document also shows that the growth team presented executives with a question: “Are we comfortable launching private-by-default for teens with the identified retention and engagement drops?” Whether any Meta staff who participated in the discussion came to a “yes” or “no” answer is not revealed in these documents. In response to questions about this document, Stone said that in September 2020, the company began developing tools that would set new accounts for teens under the age of 16 to private by default.

[Ellen Cushing: How Facebook fails 90 percent of its users]

But other contemporaneous documents show that the conflict within the company continued. In October 2020, one employee told a colleague that “private by default for teens” had been considered by the well-being team, “but the growth impact was too high and the decision was to explore more nuanced and less blunt solutions,” according to a chat transcript shown Wednesday to the jury in New Mexico. The jury was also shown the chat log about “IIC T1” content, from November 2020.

Months later, in March 2021, Meta announced a complicated solution to address these problems. Rather than making all teen accounts private by default, it would prompt new users under the age of 18 to consider making their account private, and it would ban direct messages between minor accounts and adult accounts they do not follow. A few months after that, the company went further by defaulting the accounts of 13-, 14-, and 15-year-olds to private. But it still allowed them to switch the setting back to public by themselves, without the consent of a parent, if they chose.

This update did not solve the teen-safety issue. On a single day of testing in 2022, according to another internal audit, Instagram’s “Accounts You May Follow” feature recommended, roughly 3.4 million times, teen accounts to adults who had potentially violated its policies. The same audit stated that 37 percent of users identified as “potential violators”—people who were suspected of inappropriate interaction with children—were shown at least one unconnected teen account during a test on November 25, 2022.

Yet another internal audit found that in June 2023, on a single day, 238,000 messages were sent from adults to teenagers with whom they weren’t already connected on the platform. This accounted for 9 percent of all new threads started by an adult with a teenager that day. At this point, Meta determined that the teen-privacy settings rolled out in 2021 were executed in an “inconsistent” way, according to the summary section of the document. Instagram was not always succeeding in preventing adults—including “High Risk Adults” whose accounts had “suspicious behavioral patterns”—from reaching out directly to minors. The document also noted that this was a public-relations problem: “Teens may be exposed to unsolicited contact from unknown adults,” the report continued, “which may not fully meet Meta’s external commitment.”

Meta now says these audits prompted future changes at the company and show that executives were taking the issue seriously at the time.

Instagram finally made all new accounts belonging to 16- and 17-year-olds private by default in September 2024; at this time, it also defaulted existing accounts belonging to minors under the age of 16 to private, as part of a new moderation-and-supervision scheme called “Teen Accounts.” Still, the policy had some qualifications. Even at this point, 16- and 17-year-olds could make their own accounts public, and users under 16 could still change the setting if a parent or guardian approved it.

More changes followed. In April 2025, Meta rolled out Teen Accounts on Facebook and Messenger, began to require parental permission for livestreaming on Instagram, and started testing an AI system designed to figure out when teenagers were lying about their age to get around restrictions. At the end of last year, to the irritation of the Motion Picture Association, Instagram released a new content-filtering system tied to the “PG-13” standard, which content-moderation experts found baffling.

[Read: The end of the old Instagram]

Meta often defends itself by pointing to public timelines it has published, which explain the many changes that it has made to improve safety on its platforms. For instance, in February 2023, Meta joined an effort by the National Center for Missing and Exploited Children that allowed people to report and remove online intimate photos of children under the age of 18. Later that year it joined a network of other tech companies to share information about bad actors on their platforms. Soon after, it added a feature to teens’ accounts that automatically blurs images that may contain nudity. Stone also noted that Meta automatically disables the accounts of suspicious adults who are flagged for behaviors such as being blocked by a teen or searching for particular terms associated with sexual predation.

These facts also help emphasize an important point about the case in New Mexico: The version of Instagram at issue in the lawsuit was the one that teens used in 2023, which was substantially different from the product as it exists today. Meta has also raised issues with the approach that Torrez has taken: To construct his case against Meta, Torrez coordinated a sting operation to demonstrate how children are victimized on Instagram. In 2023, his office used decoy accounts, pretending to be kids 14 years old or younger, to demonstrate that it was both easy and common for adults to contact children, to converse with them in sexually explicit terms, and to send them sexually explicit material. Meta has criticized this strategy—the company’s attorney Kevin Huff called it “rigged” to produce a “fake result.” Three men entered guilty pleas in New Mexico after attempting to meet underage girls on Meta’s platform during the sting. The New Mexico jury is currently hearing testimony to decide on the merits of the approach.

Meta has insisted that internal documents that reach the public are plucked out of context. Witnesses, including former employees who were privy to excerpted closed-door debates, will get the chance to provide context, interpreting the evidence for both a jury and a national audience. Many others will also have their day in court, as New Mexico’s case against Meta is just one of many. It will soon be joined by a federal case bundling more than 2,000 complaints of personal injury.

In the meantime, the company is doing a full marketing push for Teen Accounts. One recent ad shows Tom Brady discussing the value of screen-time limits with his son. And a previous entry shows a montage of mothers caring for their children. “You’ve always looked out for them,” a text card in the ad reads. “We’re here to do it with you.”

Marie-Rose Sheinerman and Isabel Ruehl contributed reporting.

This article has been updated to include additional information about the sting operation in New Mexico.

Can Instagram Ruin Your Life? The Jury Will Decide.

2026-02-09T21:00:00-05:00

In a state court in Los Angeles this week, 12 jurors are hearing opening arguments in a case that has the potential to change social media—maybe even the internet—as we know it.

The trial, which began today, is a bellwether: Similar individual cases have been filed all around the country, and a massive federal case with more than 2,000 plaintiffs is expected to proceed this summer. In each case, the plaintiffs accuse social-media companies of releasing defective products. The argument is that these products were built with dangerously habit-forming features—including the endless-scroll feed, algorithmic recommendations, and push notifications—that have led to an array of serious health problems. Plaintiffs also accuse the companies of failing to warn users about the risks of using their products and of deliberately concealing their dangers.

The L.A. case is the first to make it to trial. It is scheduled to last about six weeks, and it focuses heavily on Meta—in particular, Instagram. (The defendant originally included TikTok, Snap, and YouTube. TikTok and Snap settled with the plaintiff last month rather than go to trial. YouTube remains part of the case, though it is less central to the complaint. The company has said that allegations against it are “simply untrue.”) The lawsuit asks an existential question about Meta’s business: Can the fundamental design and most basic features of Instagram directly cause mental-health problems in kids and teenagers? The jury will be asked to answer that question, and Meta is taking a big risk by allowing it to do so (though it can appeal to a judge if it loses).

The plaintiff in this case is a 19-year-old California woman who is named only by her initials, K.G.M., because the events that she’s suing over happened when she was a minor. Her suit states that she began using social media at the age of 10 and alleges that her mental health was directly degraded by Instagram. According to her complaint, the app “targeted” her with “harmful and depressive content,” which led her to develop a negative body image and to commit acts of self-harm. She also says that she was a victim of “bullying and sextortion” in the app as a minor and that Instagram did not do “anything” until her friends and family spent two weeks repeatedly reporting the problem. Her older sister, a plaintiff in a separate case, suffered a life-threatening eating disorder that the family believes was also triggered by usage of Instagram and other social-media sites.

The basic allegations do not make Meta look very good. The company may be taking its chances in court now simply because it has to eventually. If it were to win this case, that might slow the momentum of all the others coming. The company may also relish an opportunity to set the record straight, as it were. For years now, Meta has been compared to Big Tobacco and accused of deliberately destroying children’s minds. Internal documents leaked by the whistleblower Frances Haugen in 2021 showing that some employees were worried about Instagram’s effects on young girls made matters worse. In response to the backlash, which has been ongoing ever since, the company has half-acquiesced to public pressure and made piecemeal efforts at image rehabilitation. It has explained itself in dry blog posts, created more ornate parental controls, and launched awkward ad campaigns emphasizing its commitment to safety and screen-life balance. (In its latest ad, Tom Brady describes his teen son’s ability to connect with friends online as “very much a value-add.”)

[Listen: The problem with comparing social media to Big Tobacco ]

Now the company will see if it can possibly sway a group of ordinary Americans with its version of the facts. “This will be their first chance to tell their story to a jury and get a sense of how well those arguments are playing,” Eric Goldman, a professor at Santa Clara University School of Law, told me. Meta’s day in court has come.

K.G.M, like many of the other plaintiffs filing personal-injury suits against social-media companies, is represented by the Seattle-based Social Media Victims Law Center. In the spring of 2023, the organization filed a complaint on behalf of a number of plaintiffs, opening with this animating statement: “American children are suffering an unprecedented mental health crisis fueled by Defendants’ addictive and dangerous social media products.”

The complaint goes on to accuse social-media companies of deliberately “borrowing” tactics from the slot-machine and cigarette industries in an effort to make their products addictive, and argues that social-media apps have “rewired” kids so that they prefer digital “likes” to genuine friendship, “mindless scrolling” to offline play. “While presented as ‘social,’ Defendants’ products have in myriad ways promoted disconnection, disassociation, and a legion of mental and physical harms,” the complaint summarizes. In K.G.M.’s case, the listed harms include “dangerous dependency” on social media as well as “anxiety, depression, self-harm, and body dysmorphia.”

Her case is the first of potentially thousands. Numerous school districts, state attorneys general, tribal nations, and individuals have also filed suit against social-media companies. But this case is worth watching because it will hit on all of the big topics. To assess whether social media is generally harmful to kids and teens, lawyers will have to argue about the nitty-gritty of a complicated and conflicted scientific field. To get at the question of whether Meta hid specific knowledge of harm, they’ll debate the meaning of the documents Haugen leaked as well as others produced during discovery.

[Read: A tiny outrage machine, sucking the exhaust from a giant one]

The jury will likely hear arguments about whether social-media addiction is real, what the murky concept of “the algorithm” actually means, and whether the richest companies in history really have allowed bad things to happen to children for the benefit of their bottom line. Reached for comment, a Meta spokesperson pointed me to an informational website the company has created about the lawsuit and highlighted a previous statement, which reads in part: “Plaintiffs’ lawyers have selectively cited Meta’s internal documents to construct a misleading narrative, suggesting our platforms have harmed teens and that Meta has prioritized growth over their well-being. These claims don’t reflect reality.”

Goldman, who often writes about internet law, said that he thinks Meta will have its work cut out for it with the jury. After 10 years of critical media coverage and political bickering about how to rein the tech companies in, “I assume that the jury is going to walk into the courtroom heavily skeptical of Facebook, Instagram, YouTube, and social media generally,” he said.

Meta’s lawyers can make a good scientific case on some of the broader questions. Researchers have looked for years for smoking-gun evidence that social-media use directly causes mental-health problems in young people at scale, and have mostly turned up weak and inconsistent correlations and no way to prove long-term causation. Major scientific bodies such as the National Academies of Sciences, Engineering, and Medicine have started to recognize that the story is more complicated than just saying that social media is dangerous in all forms and for all kids.

However, this case is about one kid. Even if social-media addiction is not “real” in the sense that it is not in the DSM-5, and even if it has not created a mental-health epidemic all on its own, certain people, perhaps many, could still be susceptible to what some clinicians prefer to call problematic internet use. The jury will have to decide whether that can cause further problems such as the ones K.G.M. has described (and whether it’s Meta’s fault if it does). Legally, the burden will be on her lawyers to convince them of that.

[Read: No one knows exactly what social media is doing to teens]

This is a sticky situation. Corbin Barthold, the internet-policy counsel at the think tank TechFreedom, told me that “having lawyers get up and give speech contests in front of a jury” is one of the worst ways he can imagine of settling the scientific disputes about social media and its effects on mental health. (Actually, he called it “crazy.”) And it is somewhat surprising that we’ve ended up here. Social-media companies are usually protected by a portion of the 1996 Communications Decency Act known as Section 230, which guarantees that online platforms are not considered legally responsible for what their users post or see. The law has been the subject of repeated controversy and legal challenge ever since it was written. Some people now argue that it is totally outdated, having been written at a time when the web was essentially a bunch of static pages, nothing like the complicated landscape we spend so much time in today.

Meta tried and failed to have the case dismissed on Section 230 grounds. Judge Carolyn Kuhl let it proceed because it will not consider specific posts or comments; instead, it will focus on design features such as the recommendation algorithm and the never-ending feed. Free-speech civil-society groups on the right and the left were irked by Kuhl’s decision. However, Kuhl is not the only judge who has recently allowed such arguments to go ahead. A similar product-liability claim was the basis of a lawsuit against Google and Character.AI, filed in 2024 by the mother of a 14-year-old boy who killed himself after forming an intense relationship with a chatbot. That case was settled out of court, but it signaled, as the University of Buffalo School of Law professor Mark Bartholomew put it to me in an email, a shift, and evidence of “a growing willingness” among the courts “to take old product liability doctrines for physical goods and apply them to software.”

This trial is just one specific personal-injury suit as well as, possibly, the first of many. “It’s a brick in a potential wall,” James Grimmelmann, a professor of digital and information law at Cornell Law School, told me. “If they think they’re going to keep on losing other cases, they’re going to have to make changes.” It’s not yet obvious what changes the company would have to make. No more content recommendations? No more feed? It’s not just Meta whose future would be in question. It would be any internet-based service that has any reason to believe that anyone under the age of 18 could be using it and getting “addicted” to it.

[Listen: The smartphone kids are not all right]

The possibly enormous stakes reflect how pitched the debate about social media has become. Pete Etchells, a professor of psychology and science communication at Bath Spa University, in England, told me that he finds the situation “really frustrating.” One side denies that anything is wrong; the other side compares social media to cigarettes, even though that makes little sense. “We’re not talking about a biological substance that you can consume that has a demonstrable chemical effect,” Etchells said.

Etchells wrote a book titled Unlocked: The Real Science of Screentime, which was published in 2024 and argued, in part, that a moral panic about social media and smartphones has been making it more difficult to learn how to use them in beneficial ways and how to pick apart what, specifically, might be wrong with them. At the same time, the public justifiably wants something done about the unaccountable tech companies, he said, and bridles when those companies seem to be cherry-picking scientific studies that fit their narrative, throwing them up as an ironclad defense in order to avoid reflection again.

Even if science is on those companies’ side in a general sense, that doesn’t necessarily mean that the facts are on their side when you talk about one girl, one series of particular events. And now, after years of hearings and reports and rebuttals and failed legislation and bad ideas and ad spots, it’s all up to that jury. They have the task of looking at this one story, hearing both sides, and making a decision.

America Will Be Reading the Epstein Files for Decades

2026-01-30T19:03:32-05:00

Updated at 8:25 p.m. on January 30, 2026

Reporters, lawmakers, and ordinary Americans are poring over a deluge of new files related to the Jeffrey Epstein case today, following the latest release from the Department of Justice. This release is substantially larger than any previous ones, with 3 million pages of documents, more than 180,000 photos, and more than 2,000 videos, according to the DOJ. The website they were uploaded to—which has the elegant URL Justice.gov/Epstein—is not intuitive to operate and offers a search box as its primary navigation tool.

But a search box was the only thing many viewers needed, as they were diving into the files in pursuit of information on specific people—President Trump in particular. Thousands of the documents appear to mention Trump, though not all of them in any significant way (for example, The New York Times notes that some of the documents are copies of news articles that contain his name). When Deputy Attorney General Todd Blanche announced the file release earlier today, he said that the White House had “nothing to do” with their vetting and “no oversight.” The DOJ press release further emphasizes the independence of the process and says that “notable individuals and politicians were not redacted in the release of any files.” The only redactions, according to the release, were those made to protect victims and their families.

Reporters sifting through the files have found plenty of news. They’ve turned up a series of emails that Epstein wrote about Bill Gates and then sent to himself in 2013. In these, Epstein suggests that he helped Gates have extramarital affairs and expresses disgust that Gates would “discard” their friendship after asking Epstein to do things “that have ranged from the morally inappropriate to the ethically unsound” and “potentially over the line into illegal.” (The Gates Foundation has already issued a comment to the Times that the claims—“from a proven, disgruntled liar—are absolutely absurd and completely false.”) Many other notable people appear in the files, including Bill Clinton, who was also in previously released photos. A number of the documents referencing Clinton are uncorroborated tips sent to the FBI. A friendly email exchange between Epstein and Elon Musk turned up, as did a reference to Kevin Warsh, Trump’s preferred candidate to replace Jerome Powell as Federal Reserve chair. (Both Bill and Hillary Clinton have refused to testify for the House Oversight Committee’s Epstein investigation, and they recently released a joint statement saying that they’ve already shared “the little information” they have about Epstein. Musk and Warsh did not immediately respond to requests for comment.)

The files about such a sensitive and complicated series of crimes could be released to the public only by way of a convoluted procedural process, but the process came to seem suspiciously convoluted to many Americans in large part because of the president’s many reversals and evasions. Before Trump returned to office, he expressed a somewhat relaxed attitude about the Epstein files. Asked on the Lex Fridman Podcast in September 2024 whether he would release them if he were reelected, he said, “Yeah, I’d be inclined to do the Epstein; I’d have no problem with it.” Early last year, to demonstrate his dedication to truth-telling and transparency, Trump invited MAGA influencers to the White House, where they received binders full of Epstein-related documents labeled Phase 1. But some of the influencers were chagrined to find that the files inside were not new.

Then, in May, Trump reportedly learned from Attorney General Pam Bondi that he himself was mentioned in the unreleased files. Two months later, his attitude about them had dramatically changed. When the FBI announced in July that it had reviewed all of its Epstein files and would not release any more after all—that there would be no Phase 2—both Trump’s enemies and the MAGA faithful were aghast. But Trump was defiant and dismissive. “This guy’s been talked about for years,” he said to a reporter. “Are people still talking about this guy, this creep? That is unbelievable.” His resistance to releasing any more information caused a schism among his supporters (and contributed to the resignation of his longtime ally Marjorie Taylor Greene) and set Congress up for an easy win. Lawmakers moved quickly to pass a law that required the release of all further documents by December 19. Now, weeks past the deadline, the file dump has arrived.

After months of rumors about why Trump so stridently demanded that everyone move on from the Epstein files (and speculation that anything else he did was an effort to distract from them), people immediately latched onto a six-page FBI memo included in the dump that, at least on the surface, seemed to lend credence to their darkest guesses. The memo contained a spreadsheet full of uncorroborated tips sent to law enforcement, many alleging Trump’s participation in violent sex crimes involving minors, and some including graphic descriptions of their allegations. The document—which has already been widely shared by politicians and political commentators—includes some brief details about how law enforcement followed up on the tips but not about whether the tips were resolved. It also includes a note that some of the reports were “second-hand information.” (Trump has downplayed his past relationship with Epstein and denied any wrongdoing.)

This spreadsheet was salacious enough on its own, but then it was briefly unavailable on the Justice Department’s website, prompting people who had already downloaded it to start re-uploading it to other sites, recirculating it on social media, and speculating that it had been deliberately pulled down for political reasons. “DOJ has since killed this link,” the reporter Jake Tapper wrote on Bluesky, sharing a screenshot of the memo. “This is what was there.” (The document is currently loading on the DOJ site without a problem.)

Reached for comment, the DOJ spokesperson Natalie Baldassarre told me that the FBI memo had been “down due to overload and is back online.” She did not immediately respond to a follow-up email asking for more detail. Baldassarre also directed me to a section of the original DOJ press release about these documents. It notes that everything that the FBI received from the public was covered under the scope of law and had to be released, even if the information was unvetted. “Some of the documents contain untrue and sensationalist claims against President Trump that were submitted to the FBI right before the 2020 election,” the press release continues. “To be clear, the claims are unfounded and false, and if they have a shred of credibility, they certainly would have been weaponized against President Trump already.” Very neutral wording. (The White House’s press team referred me to this same release when I contacted them for comment.)

But the public is unlikely to look away, even if the DOJ suggests that it should. The upswell of outrage over the botched web-hosting is a repeat of the suspicion and testiness that surrounded the issue last summer, when the FBI tried to discourage theorizing about a rumored “client list” and the footage from Epstein’s jail cell. Americans have been roped even further into this dark, tragic story because of the president’s bizarre equivocation and emotional outbursts about it. They’ll be going through these files for a long time.

The New Shadowbanning Panic

2026-01-28T18:00:00-05:00

Over the past several days, TikTok users have found themselves at a loss. Literally, I mean: They lost their audiences, and their view counts showed “0.” Some people who attempted to upload content about anti-ICE protests or the killing of Alex Pretti alleged that the platform was intentionally blocking them from doing so. Others were able to get their videos uploaded, but alleged that TikTok was not distributing them. Still others noticed that they were unable to send the word Epstein in a direct message, a quirk so bizarre that it incited California Governor Gavin Newsom to repost a screenshot shared by an anonymous X account using the handle @intelligentpawg.

For many of these people, the explanation was obvious: “MAGA censorship.” Newsom said in his post that he would be launching a review “into whether TikTok is violating state law by censoring Trump-critical content,” and the concern isn’t totally random. TikTok’s U.S. business changed hands just last week, spinning off from the Chinese company ByteDance—as required by a 2024 law—and into a new organization called the TikTok USDS Joint Venture LLC. Major investors in the new entity include Oracle, which was co-founded by Larry Ellison, a Trump ally whose son is in charge of one of the country’s largest media conglomerates and is also cozy—or attempting to be cozy—with the president. Is it so far-fetched to imagine they would tweak the platform in his favor?

Jamie Favazza, a spokesperson for TikTok USDS Joint Venture, wrote to me in an email that the app is being run as it was before and that American users will “continue to have the same experience they already know and love.” The company has blamed the issues on a power outage at one of its data centers, which it said caused a “cascading” systems failure affecting all types of content (not just posts about Minnesota). The data center in question is operated by Oracle, which has managed TikTok’s U.S. user data since 2022. The company was still working to fix some of the bugs as of today. Separately, the company said that there is no rule against saying Epstein and that this glitch was caused by a technical issue with its safety systems. (When I sent the word Epstein to my fiancé at 8 yesterday morning, it went through fine.)

Unsurprisingly, a lot of people are not totally buying the explanation. Trump has joked that he would make the app’s algorithm “100 percent MAGA” if he could, and it’s true that aspects of the broader media ecosystem have shifted dramatically in Trump’s favor: starting with Elon Musk’s takeover of X, then major TV networks’ capitulation to Trump, and now TikTok’s transfer to Trump-friendly investors. And Trump has never been shy about applying pressure to companies in order to satisfy his own whims.

Then factor in the stakes of the moment. As federal agents threaten the basic principles of democracy in Minnesota, Americans are looking to their phones for up-to-date and on-the-ground information about a complicated, ongoing event. For better or worse, millions of Americans use TikTok for news: Any whisper of intervention or suppression is natural cause for concern. (With everything else going on, people haven’t forgotten about the Trump administration’s bizarre handling of the Jeffrey Epstein files.)

[Jonathan Rauch: Yes, it’s fascism]

That’s a perfect storm for paranoia about TikTok’s actions, and the platform is not aided by recent history. In 2020, the company apologized in response to an outcry about inexplicably low view counts on videos about Black Lives Matter. It also cited a technical glitch in that case, and some people pointed out that other popular, politically neutral hashtags (such as #cat) were affected, too, but suspicion lingered. More recently, some American TikTok users have felt directly censored by their government—leading up to the legislation that forced the app’s sale, several lawmakers blamed TikTok for, in their view, warping the minds of young people and making them overly critical of Israel, and cited this as a reason to regulate it.

Feed-based, view-based social-media platforms are central to American political discourse, which is why politicians so often fight over the details of their operation. In recent memory, it was more often Republicans calling for investigations of platform censorship, shadowbanning, and collusion between the White House and social-media companies. Back then, efforts by researchers to study the problem generally found that there was no blanket, pervasive bias against right-leaning viewpoints per se, although right-wing users were more likely to spread certain types of low-quality content, namely misinformation, which made them more likely to be penalized and therefore affirmed their feeling of being silenced.

Now we’re in a counterintuitive and paradoxical stage of content moderation, where some spaces are more chaotic than ever and others are more restricted in highly specific ways. On the one hand, you have Musk’s X, which has removed most guardrails from public discourse, up to the point of enabling users to generate nude images of their political foes; on the other, you have Meta’s Instagram, which has been contorting its rules in response to pressure from parent groups and politicians to make the app safer for teens. The interface now accuses users of looking for child-sex-abuse material if they search the phrase hot girls.

Into the breach steps a celebrity who says she is being prevented from posting about “🧊” or a journalist who claims that “the new TikTok algorithm has ZERO, and I mean absolutely ZERO news or politics content, not one word about anything going on at all, not even the weather.” In the replies to the latter post, others pushed back a bit, saying that they had actually seen plenty about the Minnesota protests and that the app was just performing weirdly in general. But the debate is unresolvable, because users have no objective way to assuage their own doubt or confirm their own fears. (I’ve also seen people start to question the visibility of political content they’ve posted on Instagram, despite the obvious fact that a ton of similar content has been highly visible there.)

Three years ago, after Musk acquired Twitter and before he turned it into X, he went through a phase of personally investigating users’ claims that they had been shadowbanned by the prior ownership. At the time, I wrote that this wouldn’t eliminate anxiety about the platform’s secret machinations. For that story, I spoke with Laura Savolainen of the University of Helsinki, then a doctoral student and now a postdoctoral researcher, about how hard it is to drag people away from the folk theories that they come up with about how the algorithm is treating them and why their content is or isn’t being seen as widely as they think it should be. “Algorithms are very conducive to folklore because the systems are so opaque,” she said then.

The reason to be paranoid about platform censorship is always the same—whether it’s happening or not, it could happen. When people feel especially reliant on social-media platforms not for stimuli, shopping, or slop, but for vital information and feelings of cohesion, support, and action, the possibility is never more real.

The Sports Conspiracy That’s Too Easy to Believe

2026-01-23T12:01:11-05:00

The San Francisco 49ers will not be playing in the Super Bowl, because they lost to the Seattle Seahawks by a disgraceful score of 41 to 6 over the weekend. But of course, “someone wins, someone loses” is never the whole story in sports.

Some fans are now buying into the narrative that the team had no chance because it suffered a suspicious number of devastating injuries over the course of the season (as well as in recent past seasons). Any fan could rattle off a list—quarterback Brock Purdy has dealt with several injuries (coming off a torn UCL a couple of years ago), three key players tore their ACLs this season, and one also tore his MCL. While the team was getting demolished by the Seahawks, tight end Jake Tonges went down with a plantar-fascia injury. To explain this, some fans pointed to the fact that the team’s practice field and stadium are near an electrical substation and suggested that the electromagnetic waves emanating from it could be weakening players’ bodies, making them especially susceptible to soft-tissue injuries such as tendon tears and muscle strains.

“The 49ers absolutely should spend whatever it takes ($20 million?) to move the electrical substation away from their practice facility,” one viral post suggested during the game. Reporters are now asking team leadership about the issue; on Wednesday afternoon, the team’s general manager, John Lynch, said in a press conference that he’s actively looking into the situation. “We’ve been reaching out to anyone and everyone to see Does a study exist?” he said.

To be clear, the concerns are baseless. The United States has thousands of electrical substations; people live near them, and other sports teams practice next to them, too. Some scientists say that you can never totally rule out the possibility that this type of ambient electromagnetic exposure might cause harm, but they have also looked for decades without producing any convincing evidence of such harm. Geoffrey Kabat, an epidemiologist and the author of 2008’s Hyping Health Risks: Environmental Hazards in Daily Life and the Science of Epidemiology, was part of a group of scientists looking for a possible association between electromagnetic-field exposure and breast cancer in the 1990s. They didn’t find any. “We’re dealing with a very weak form of energy,” he told me. There is just no proven way for these weak fields to disrupt human cells, so long as their frequencies stay below the threshold that causes tissue heating (as happens in a microwave).

But fans believe what they believe. They are obligated to live in a state of unceasing anxiety, superstition, and fear of their team being cursed. The power-plant theory is an iteration of that tradition—a movement away from mystical explanations such as the Curse of the Billy Goat and toward shareable deep dives, investigations, and dot-connecting. With all the tools available to them in the digital age (reams of hard data, unlimited multimedia “evidence”), fans can explain their teams’ worst turns of bad luck by composing what can only be called conspiracy theories. Today’s fans will go down YouTube rabbit holes; conduct amateur, open-source investigations; and cut numbers this way and that until they reveal a startling pattern.

[Read: The secret joy of baseball curses]

This particular theory is emerging during a time when alternative theories of science and medicine are central to American culture—Donald Trump’s EPA has indulged internet-y concern about chemtrails, and we’re back to bickering about conventional wisdom on some of the most basic questions of health and safety, including whether milk ought to be pasteurized. Why not take another look at electricity too, while we’re at it?

The substation theory seems to have been percolating for a while, but it blew up on social media in early January after a holistic-health influencer named Peter Cowan posted a lengthy, scientific-sounding explanation of how “low-frequency electromagnetic fields can degrade collagen, weaken tendons, and cause soft-tissue damage at levels regulators call ‘safe.’” He referred to the 49ers as a “real world case study,” and a faction of 49ers fans latched on. Chase Senior, a popular online commentator within 49ers fandom, shared parts of Cowan’s theory on X (leaving out some of its odder digressions, such as one about the Cold War), and contributed significantly to its spread.

“When it gets to the point where the team is so injured every single season, I think that it is fair to look into anything and everything to try to determine why they’ve been so hurt,” Senior told me. Initially, most people were hostile to the idea. They called him, he said, an “idiot.” Some people still call him that, and theorize that the injuries likely have something to do with the style of the team’s play or its approach to strength and conditioning or stretching and recovery. Many also pointed out that the 49ers are fairly old, as a team, and that they played on short rest a few times this season.

Indeed, the most obvious explanation for the 49ers’ problem is that football is a violent sport. The NFL has an injury epidemic, Matt Maiocco, a 49ers beat reporter for NBC Sports Bay Area, reminded me when I called to ask about the substation theory. “There’s three kinds of football players,” he said. “Guys who have been injured, guys who are already injured, and guys who will be injured.” The team has practiced in the same spot in Santa Clara for decades, he pointed out, and the substation has been there for much of that time. I asked him whether the 49ers really have had particularly bad luck with injuries the past few years. “It feels that way,” he said. “And I think the closer you are to something, the more it feels that way.” (Sports bloggers who crunch the numbers find that they are among the more injured teams—but to reiterate, all football teams are very injured.)

None of these explanations is very compelling to Senior, who has continued bringing the conspiracy theory up online, with increasing defensiveness. He says he noticed a shift during the postseason. “The Niners continued to get hurt,” he said. During the first round of the playoffs, on January 11, the 49ers beat the defending Super Bowl champions, the Philadelphia Eagles. But tight end George Kittle, one of their best players, went down early in the game with a torn Achilles tendon—the type of devastating injury that happens out of nowhere. With that, more fans and commentators grew intrigued, and others started to couch their disapproval of the theory. Even if odds are low that there’s anything to it, is it so hard for someone to check it out?

Some current and former football players have started talking about the substation theory and giving it credibility. The 49ers wide receiver Kendrick Bourne joked about the situation himself after the game against the Eagles. A reporter asked why his team was suffering so many injuries. “Yeah, it’s that power plant,” he said. “Nah, I’m just playing, I don’t know.” He went on with a more typical professional-athlete answer along the lines of how injuries are really unfortunate, but you have to have a next-man-up mentality and keep going, etc.

Now that a pattern has been proposed, it will become only more visible. Each time a 49er is injured, the power plant will come up again (mirroring the endless repeatability of the #DiedSuddenly anti-vaccine conspiracy theory, which was, for a time, invoked whenever someone died unexpectedly). But it also isn’t that serious. The reporter who asked Lynch about the substation seemed embarrassed to be talking about it, and most reasonable people clearly understand that the idea is far-fetched. Some fans are having fun with it and indulging in a little gallows humor to help them get over a disappointing end to their season.

The electromagnetic waves aren’t powerful enough to hurt an NFL player—the real question is whether the story about them is powerful enough to become a durable piece of 49ers lore. When I asked a friend of mine who grew up as a 49ers fan about the substation, she waved it off and proposed an alternative theory for the heartbreaking losses and horrifically timed injuries of the past 10 years. She said that she didn’t know anything about EMFs, but that this was something all true fans believe: The 49ers should never have left Candlestick Park, their beloved old home in San Francisco, where they reigned in their glory days, to go to Levi’s Stadium, their bland, uninspiring new one in Santa Clara, where they have seen nothing but suffering. When they did that, they offended the universe, the Earth, or God. That was serious.

Oh, Great, Another Supermoon

2026-01-03T06:45:00-05:00

For the past several years, I’ve been experiencing a tension in my relationship with the moon. I love the moon as much as anyone, but the problem, bluntly, is that the moon is too famous. Maybe you’ve noticed this. The moon is constantly in the news. It is doing something “rare” or “unique” seemingly every week. Local-news outlets will inform their readers that a supermoon is about to “take to the skies” or rise “over Milwaukee,” in stories that are not technically inaccurate, though they do fail to acknowledge that the moon is always taking to the skies and that it rises over everyone. (They will often also give advice on how best to view the moon, as though most of us don’t know generally where it is.)

National outlets do the same thing. The main difference is that Newsweek will claim that a supermoon is rising not over Milwaukee but over the United States. A partial list of outlets that covered the supermoon last month includes Time, Mashable, Live Science, PBS, ABC News, Wired, CNN, Vogue, CNET, USA Today, the Associated Press, and The Washington Post. Forbes chose to innovate by referring to it as a Christmas supermoon, even though it was visible only for a short period in the first week of December. Elle also had a creative take, which was that “supermoon season” would have some kind of profound effect on our minds and bodies (as is generally the idea when moon coverage intersects with astrology).

[Read: No one actually knows what a moon is]

Admittedly, this has become a bit of a fixation for me. I’ve tried to get my colleagues worked up about it too, a few times, by sharing links and writing “STOP TALKING ABOUT THE MOON!!!” in a Slack channel. The December supermoon coverage led me to notice—though only with a huh, look at that—the December supermoon, which was pretty bright and reminded me again of this interest. So, as the year drew to a close, I thought I might as well ask and answer the question “Why is everyone talking about the moon all the time?”

On some level, it’s obvious. The moon is great. Many moon events have tantalizing names. A Blood Moon, for instance, is another name for a total lunar eclipse, which happens every few years. A supermoon is another name for a full moon that occurs when the moon is at the point in its orbit that brings it closest to the Earth, providing the illusion of a bigger, brighter moon, which usually happens three or four times a year. Depending on when the supermoon occurs, it has a different name—a March supermoon is a Worm Moon; a May supermoon is a Flower Moon. In December we had a Cold Moon, which doesn’t look very different from other supermoons, but has a different name because it happens when it is cold. These names tie us to our forebears, in this case by reminding us of older ways of keeping time.

The names can sometimes be confusing. A Blue Moon—familiar from “once in a blue moon,” a phrase indicating extreme infrequency—is the term for when a second full moon appears in one calendar month, which happens every two or three years. It’s possible for a moon to be super and blue. It’s also possible for a moon to be super and blood, as was the case with the Super Blood Moon in September 2015, when all of this, apparently, started.

The Newspapers.com archive shows essentially zero interest in supermoons before 2010, then a small spike in coverage in 2011 around a March supermoon that was somewhat spooky looking, and then an enormous spike in the fall of 2015, which happens to have been right in the middle of the digital-media boom, when newer journalism companies were obsessed with pageviews, shares, and time spent on site. At The Verge, where I worked at the time, we ran this headline: “Tonight, a Supermoon Will Shine Red With the Blood of the Innocent.” That same moon was, as one of BuzzFeed’s several stories on the event put it, “Big and Red AF.”

Jeff Jarvis is a writer and emeritus journalism professor who has been critical of the traffic-chasing business model. When I asked him what he made of moon news, he said he’d wondered about it himself recently. He assumed it was part of a tradition going back to the days of “scissors editors” in newsrooms—folks whose whole jobs consisted of cutting stories out of competing newspapers so that they could be copied. The internet only makes this copying process go more quickly, Jarvis told me. This reminded me that in 2015, many newsrooms would look at a now-defunct online tool called CrowdTangle to see what kinds of stories were performing the best for other outlets. Primarily, what you could see was what was being shared the most on Facebook and other social platforms, including Reddit, which was in itself a good place to source viral stories because of the “Hot” list on its homepage.

[Read: The moon is leaving us]

Jarvis compared the moon to the famous Dress. In 2015, a BuzzFeed staffer saw a photo of a dress on Tumblr, which she copied into an article under the headline “What Colors Are This Dress?” There was some kind of optical illusion happening: Many people saw it as blue and black; others as white and gold. The post, which included a poll that users could vote in, was a huge hit for BuzzFeed, and for other websites that aggregated it. The moon is also a naturally occurring, free source of traffic. And Ben Smith, who was the editor in chief of BuzzFeed News in 2015 and wrote a book about the golden age of traffic (called Traffic) in 2023, also likened the moon to the dress. “This feels like the last gasp of the old, good, universalizing internet,” he wrote to me in an email. He called the moon the “platonic ideal” of “Is this dress blue or white?”—in other words, the platonic ideal of shareable content.

The difference is that the dress (73 million pageviews) happened only once. “The beauty of the moon is it keeps coming back,” Jarvis said. Writers can write about the moon doing what the moon does again and again, for the rest of time. “It’s a waste and stupid, but it’s harmless,” he said. “Nobody gets offended by the supermoon.” They don’t, that’s true. But I told him I have been a little offended by the cynicism—by calling on the huge and wonderful moon to serve such a small and silly purpose as generating clicks. He seemed to sympathize. He agreed that a case could be made that these stories are somehow cheapening the cosmos. “You’re hyping the moon. It doesn’t need any hype.”

Exactly. And that means the inverse is also true. Who cares if we hype the moon? The moon is unaffected. The moon is the moon forever. Our hype glances off it and does less than the tiniest meteoroid. That sturdiness and predictability are exactly why we turn to it so often in this desperate business. Internet traffic is “very mysterious,” Caitlin Petre, an associate professor of journalism and media studies at Rutgers, told me when I called her to talk about the important question of whether it’s okay to exploit the moon for our petty ends. Even with the advanced metrics that most newsrooms have access to now, you often end up guessing about the desires and interests of an undefined “audience.” But the moon is the rare topic about which there is no guessing. People love it.

Petre, the author of the 2021 book All the News That’s Fit to Click: How Metrics Are Transforming the Work of Journalists, pointed me to a famous 2011 study that found that stories evoking anger, anxiety, or awe were shared more often than other stories. Of those three emotions, awe is clearly the most ethical to try to elicit. “I guess I would say, in the annals of all the things that news organizations do to chase traffic, writing about the moon is probably one of the best ones,” she said. “That would be my take.”

So here I am writing myself into the annals of moon-hype traffic. Tonight there will be a supermoon called a Wolf Moon. You can learn about it in many ways, including via a USA Today story syndicated on many local-news websites, which I found because of the flawlessly search-optimized headline “When Is the Next Full Moon? Wolf Moon Will Be First, Bright Supermoon of 2026.” Many things are uncertain, but I know the supermoon will be beautiful and I know what it will do—it will rise over your city, wherever you are.

*Illustration by Anna Ruch / The Atlantic. Sources: John Adams Whipple / James Wallace Black / Heritage Images / Getty; Alfred Stieglitz / Heritage Images / Getty; SSPL / Getty.

Australia’s Grand Social-Media Experiment

2025-12-18T07:30:00-05:00

To celebrate the start of a nationwide ban on social media for kids under the age of 16, the Australian government lit the Sydney Harbour Bridge with the slogan Let Them Be Kids.

As of December 10, younger teenagers in Australia can no longer make accounts on popular social-media sites, including Instagram, TikTok, YouTube, Reddit, and Twitch. The minister for communications’ rule for the ban defines a social-media site as one that primarily exists to encourage interaction among users and allow them to post their own content. (By this definition, so far, Pinterest, the super-popular chat site Discord, and the online game Roblox, though they have social features, are not included.) Social-media companies are required to make “reasonable” efforts to keep people under 16 off of their apps, and they face hefty financial penalties for noncompliance.

The government’s argument for the ban has been clear: Getting kids off of social media will make them healthier and happier. Explaining the law in a June speech, eSafety Commissioner Julie Inman Grant mentioned many of the things parents worry about their kids encountering online—grooming, cyberbullying, graphic violence, sexualized chatbots, deepfake revenge porn. She also spoke to a more general dread about what social media may be doing to young people. Parents, she said, have been rightly concerned about “algorithmic manipulation” and “predatory design features” that “encourage compulsive usage.”

[Listen: Is this the end of kids on social media?]

Australia is the first country to take such sweeping action, but many countries, including the United States, are considering age-gating social media in similar ways. “We know the world will be watching,” Prime Minister Anthony Albanese said in a September speech at the United Nations.

That’s true. But what will other countries see? The problems that the Australian government is trying to fix are well recognized: Parents and educators worry that social media pulls kids away from schoolwork, outdoor play, sleep, and their friends, while making them more susceptible to various dangers, including anxiety and depression. Less clear, though, is how Australia will know whether the ban works.

Inman Grant’s office has appointed a scientific advisory group that will evaluate the effectiveness of the ban over the next two years. It will be led by Jeff Hancock, the head of Stanford’s Social Media Lab, but its specific plans for studying the ban have not been released. Susan Sawyer, a professor in the pediatrics department at the University of Melbourne and an adolescent-health researcher at Murdoch Children’s Research Institute, is part of the group, and she told me that it will likely be a few months before details will be made public.

Sawyer said that she was glad the government was investing in a robust evaluation of the ban. She also acknowledged that the task may be daunting. “I have very publicly described this as a social experiment,” she told me. “Children’s current exposure to social media is a social experiment, and the response that the Australian government is taking is a further social experiment.”

The challenge is how many variables are at play. Researching the ban’s effects will not be as easy as having the teens of the recent past serve as a control group for the teens of the near future. There are not even clear bounds around those groups. Think about the 15-year-olds in Australia right now. Some of them have already spent a lot of time on social media and are now being forced to take a break for a short number of months before their birthday. Some of them weren’t on social media before. Some of them will circumvent the ban and stay on social media.

I asked a few researchers not involved with the advisory group how they would start if they were given the task of studying the ban’s effects. Everyone I spoke with told me that the first thing they would want to know is whether the ban succeeds at the basic goal of getting kids off of social media. (The eSafety commissioner’s office has promised data on this before Christmas.) Social-media companies have been allowed to devise their own age-verification strategies, and X has reportedly entrusted the task to Grok, its, uh, mercurial AI. Kids who are motivated to get around age filters may come up with many clever and technologically savvy ways to do so, or they may just ask their parents for help. Although recent polling indicates that the law is supported by a majority of Australians, fewer than a third of parents said they would fully enforce the ban with their own kids.

The next step would be looking for changes. “Some things you can measure pretty easily in the short term,” Jeff Niederdeppe, a communication professor at Cornell specializing in public health, told me. If kids’ sleep patterns became different, or if they started spending more time doing the things adults want them to do—going outside, hanging out with friends in person, studying—those would be relatively straightforward to track. Other changes would be harder to measure, he said. Suicides, for insance, are statistically rare, so correlating them with other trends is notoriously difficult. Anxiety, depression, and even school performance will take much longer to study, and it will be harder to tie them directly to the ban even if they do change.

Niederdeppe wondered whether data the government collects will be tied to individuals—whether you’ll be able to compare Kid A’s well-being at Time 1 and Time 2—or whether they will have to be analyzed at the group level. Ideally, he said, you’d want evidence that the kids who were using social media the most before the ban were the ones who saw the most change in their behavior or health afterward. He was also unsure how researchers might go about comparing Australian teens with some kind of control group, saying that country-to-country comparisons are imperfect. “What’s a comparable group to Australia?” he asked.

Candice Odgers, a psychology professor at UC Irvine, also mentioned country-to-country comparisons as a problem. Odgers has studied several of the big questions around kids and social media, and has often challenged the idea that there is a scientific basis for the current level of concern. “Kids in Norway differ from kids in Australia for all kinds of reasons,” she told me. “I hope that’s not the pathway people go, but I fear it will be.”

If the ban were to have the major positive impacts that many hope it will, proving it would take years. Odgers believes there is a risk that people won’t wait. She worries that politicians and other adults are going to “declare victory based on anecdotal evidence alone,” the way some have with recent phone bans in American schools. Anecdotes still count for something, but both she and Erin O’Connor, an education professor at NYU, also mentioned the need to look for unintended effects of the ban—a boomerang situation where forbidding social media makes it more alluring or causes more conflict and distrust between kids and parents, or a mix of effects, where it helps some groups while harming others, such as LGBTQ teens who may rely on social media for types of connection they lack in their daily life.

Many people might reasonably argue that the dangers of social media are so apparent that it becomes pedantic, even irresponsible, to delay action while waiting for scientists to settle their arguments. Yet the Australian government has presented the case for its ban as evidence-based. The tension is that the scientific evidence of a pervasive public-health problem caused directly by social media is nowhere near as strong as the popular feeling about its obviousness. Those dangers are difficult to evaluate clearly now; it will only be that much more difficult to evaluate how well they’re combatted by any one intervention.

When Sawyer, of the eCommissioner’s advisory group, explained her understanding of the science around kids’ mental health and social media, she said that the effect sizes that most researchers find are “modest,” although she cautioned that most data are, by nature, out of date by the time they’re published and that the effects could have gotten stronger as the internet has continued to evolve in bizarre ways. Still, she added, she would not say that there is evidence that social media has caused a public-health crisis, and she would not guess that the ban could actually fix it. “I would not be suggesting that any ban on social media is going to be a silver bullet,” she said. “It’s potentially part of a solution, but only part.”

The Strange Disappearance of an Anti-AI Activist

2025-12-04T16:41:00-05:00

Before Sam Kirchner vanished, before the San Francisco Police Department began to warn that he could be armed and dangerous, before OpenAI locked down its offices over the potential threat, those who encountered him saw him as an ordinary, if ardent, activist.

Phoebe Thomas Sorgen met Kirchner a few months ago at Travis Air Force Base, northeast of San Francisco, at a protest against immigration policy and U.S. military aid to Israel. Sorgen, a longtime activist whose first protests were against the Vietnam War, was going to block an entrance to the base with six other older women. Kirchner, 27 years old, was there with a couple of other members of a new group called Stop AI, and they all agreed to go along to record video on their phones in case of a confrontation with the police.

“They were mainly there, I believe, to recruit people who might be willing to engage in nonviolent civil disobedience, which they see as the key to stopping super AI”—a method Sorgen thought was really smart, she told me. Afterward, she started going to Stop AI’s weekly meetings in Berkeley and learning about the artificial-intelligence industry, adopting the activist group’s cause as one of her own. She was impressed by Kirchner and the other leaders, who struck her as passionate and well informed. They’d done their research on AI and on protest movements; they knew what they were talking about and what to do. “They were committed to nonviolence on the merits as well as strategically,” she said.

They followed a typical activist playbook. They passed out flyers and served pizza and beer at a T-shirt-making party. They organized monthly demonstrations and debated various ideas for publicity stunts. Stop AI, which calls for a permanent global ban on the development of artificial superintelligence, has always been a little more radical—more open to offending, its members clearly willing to get arrested—than some of the other groups protesting the development of artificial general intelligence, but Sorgen told me that the leaders were also clear, at every turn, that violence was not morally acceptable or part of a winning strategy. (“That’s the empire’s game, violence,” she noted. “We can’t compete on that level even if we wanted to.”) Organizers who gathered in a Stop AI Signal chat were given only one warning for musing or even joking about violent actions. After that, they would be banned.

Kirchner, who’d moved to San Francisco from Seattle and co-founded Stop AI there last year, publicly expressed his own commitment to nonviolence many times, and friends and allies say they believed him. Yet they also say he could be hotheaded and dogmatic, that he seemed to be suffering under the strain of his belief that the creation of smarter-than-human AI was imminent and that it would almost certainly lead to the end of all human life. He often talked about the possibility that AI could kill his sister, and he seemed to be motivated by this fear.

“I did perceive an intensity,” Sorgen said. She sometimes talked with Kirchner about toning it down and taking a breath, for the good of Stop AI, which would need mass support. But she was empathetic, having had her own experience with protesting against nuclear proliferation as a young woman and sinking into a deep depression when she was met with indifference. “It’s very stressful to contemplate the end of our species—to realize that that is quite likely. That can be difficult emotionally.”

Whatever the exact reason or the precise triggering event, Kirchner appears to have recently lost faith in the strategy of nonviolence, at least briefly. This alleged moment of crisis led to his expulsion from Stop AI, to a series of 911 calls placed by his compatriots, and, apparently, to his disappearance. His friends say they have been looking for him every day, but nearly two weeks have gone by with no sign of him.

Although Kirchner’s true intentions are impossible to know at this point, and his story remains hazy, the rough outline has been enough to inspire worried conversation about the AI-safety movement as a whole. Experts disagree about the existential risk of AI, and some people think the idea of superintelligent AI destroying all human life is barely more than a fantasy, whereas to others it is practically inevitable. “He had the weight of the world on his shoulders,” Wynd Kaufmyn, one of Stop AI’s core organizers, told me of Kirchner. What might you do if you truly felt that way?

“I am no longer part of Stop AI,” Kirchner posted to X just before 4 a.m. Pacific time on Friday, November 21. Later that day, OpenAI put its San Francisco offices on lockdown, as reported by Wired, telling employees that it had received information indicating that Kirchner had “expressed interest in causing physical harm to OpenAI employees.”

The problem started the previous Sunday, according to both Kaufmyn and Matthew Hall, Stop AI’s recently elected leader, who goes by Yakko. At a planning meeting, Kirchner got into a disagreement with the others about the wording of some messaging for an upcoming demonstration—he was so upset, Kaufmyn and Hall told me, that the meeting totally devolved and Kirchner left, saying that he would proceed with his idea on his own. Later that evening, he allegedly confronted Yakko and demanded access to Stop AI funds. “I was concerned, given his demeanor, what he might use that money on,” Yakko told me. When he refused to give Kirchner the money, he said, Kirchner punched him several times in the head. Kaufmyn was not present during the alleged assault, but she went to the hospital with Yakko, who was examined for a concussion, according to both of them. (Yakko also shared his emergency-room-discharge form with me. I was unable to reach Kirchner for comment.)

On Monday morning, according to Yakko, Kirchner was apologetic but seemed conflicted. He expressed that he was exasperated by how slowly the movement was going and that he didn’t think nonviolence was working. “I believe his exact words were: ‘The nonviolence ship has sailed for me,’” Yakko said. Yakko and Kaufmyn told me that Stop AI members called the SFPD at this point to express some concern about what Kirchner might do but that nothing came of the call.

After that, for a few days, Stop AI dealt with the issue privately. Kirchner could no longer be part of Stop AI because of the alleged violent confrontation, but the situation appeared manageable. Members of the group became newly concerned when Kirchner didn’t show at a scheduled court hearing related to his February arrest for blocking doors at an OpenAI office. They went to Kirchner’s apartment in West Oakland and found it unlocked and empty, at which point they felt obligated to notify the police again and to also notify various AI companies that they didn’t know where Kirchner was and that there was some possibility that he could be dangerous.

Both Kaufmyn and Sorgen suspect that Kirchner is likely camping somewhere—he took his bicycle with him but left behind other belongings, including his laptop and phone. They imagine he’s feeling wounded and betrayed, and maybe fearful of the consequences of his alleged meltdown. Yakko told me that he wasn’t sure about Kirchner’s state of mind but that he didn’t believe that Kirchner had access to funds that would enable him to act on his alleged suggestions of violence. Remmelt Ellen, an adviser to Stop AI, told me that he was concerned about Kirchner’s safety, especially if he is experiencing a mental-health crisis.

Almost two weeks into his disappearance, Kirchner’s situation has grown worse. The San Francisco Standard recently reported on an internal bulletin circulated within the SFPD on November 21, which cited two callers who warned that Kirchner had specifically threatened to buy high-powered weapons and to kill people at OpenAI. Both Kaufmyn and Yakko told me that they were confused by the report. “As far as I know, Sam made no direct threats to OpenAI or anyone else,” Yakko said. From his perspective, the likelihood that Kirchner was dangerous was low, but the group didn’t want to take any chances. (A representative from the SFPD declined to comment on the bulletin; OpenAI did not return a request for comment.)

The reaction from the broader AI-safety movement was fast and consistent. Many disavowed violence. One group, PauseAI, a much larger AI-safety activist group than Stop AI, specifically disavowed Kirchner. PauseAI is notably staid—it includes property damage in its definition of violence, for instance, and doesn’t allow volunteers to do anything illegal or disruptive, such as chain themselves to doors, barricade gates, and otherwise trespass or interfere with the operations of AI companies. “The kind of protests we do are people standing at the same place and maybe speaking a message,” the group’s CEO, Maxime Fournes, told me, “but not preventing people from going to work or blocking the streets.”

This is one of the reasons that Stop AI was founded in the first place. Kirchner and others, who’d met in the PauseAI Discord server, thought that that genteel approach was insufficient. Instead, Stop AI situated itself in a tradition of more confrontational protest, consulting Gene Sharp’s 1973 classic, The Methods of Nonviolent Action, which includes such tactics as sit-ins, “nonviolent obstruction,” and “seeking imprisonment.”

In its early stages, the movement against unaccountable AI development has had to face the same questions as any other burgeoning social movement: How do you win broad support? How can you be palatable and appealing while also being sufficiently pointed, extreme enough to get attention but not so much that you sabotage yourself? If the stakes are as high as you say they are, how do you act like it?

Michaël Trazzi, an activist who went on a hunger strike outside Google DeepMind’s London headquarters in September, also believes that AI could lead to human extinction. He told me that he believes that people can do things that are extreme enough to “show we are in an emergency” while still being nonviolent and nondisruptive. (PauseAI also discourages its members from doing hunger strikes.)

The biggest difference between PauseAI and Stop AI is the one implied in their names. PauseAI advocates for a pause in superintelligent-AI development until it can proceed safely, or in “alignment” with democratically decided ideal outcomes. Stop AI’s position is that this kind of alignment is a fantasy and that AI should not be allowed to progress further toward superhuman intelligence than it already has. For that reason, their rhetoric differs as much as their tactics. “You should not hear official PauseAI channels saying things like ‘We will all die with complete certainty,’” Fournes said. By contrast, Stop AI has opted for very blunt messaging. Announcing plans to barricade the doors of an OpenAI office in San Francisco last October, organizers sent out a press release that read, in part, “OpenAI is trying to build something smarter than humans and it is going to kill us all!” More recently, the group promoted another protest with a digital flyer saying, “Close OpenAI or We’re All Gonna Die!”

Jonathan Kallay, a 47-year-old activist who is not based in San Francisco but who participates in a Stop AI Discord server with just under 400 people in it, told me that Stop AI is a “large and diverse group of people” who are concerned about AI for a variety of reasons—job loss, environmental impact, creative-property rights, and so on. Not all of them fear the imminent end of the world. But they have all signed up for a version of the movement that puts that possibility front and center.

Yakko, who joined Stop AI earlier this year, was elected the group’s new leader on October 28. That he and others in Stop AI were not completely on board with the gloomy messaging that Kirchner favored was one of the causes of the falling out, he told me: “I think that made him feel betrayed and scared.”

Going forward, Yakko said, Stop AI will be focused on a more hopeful message and will try to emphasize that an alternate future is still possible “rather than just trying to scare people, even if the truth is scary.” One of his ideas is to help organize a global general strike (and to do so before AI takes a large enough share of human jobs that it’s too late for withholding labor to have any impact).

Stop AI is not the only group considering and reconsidering how to talk about the problem. These debates over rhetoric and tactics have been taking place in an insular cultural enclave where forum threads come to vivid life. Sometimes, it can be hard to keep track of who’s on whose side. For instance, Stop AI might seem a natural ally of Eliezer Yudkowsky, a famous AI doomer whose recent book, If Anyone Builds It, Everyone Dies, co-authored with Nate Soares, predicts human extinction in its title. But they are actually at odds. (Through a representative, Yudkowsky declined to comment for this article.)

Émile P. Torres, a philosopher and historian who was friendly with Kirchner and attended a Stop AI protest this summer, has criticized Yudkowsky for engaging in a thought exercise about how many people it would be ethical to let die in order to prevent a superintelligent AI from taking over the world. He also tried to persuade Kirchner and other Stop AI leaders to take a more delicate approach to talking about human extinction as a likely outcome of advanced AI development, because he thinks that this kind of rhetoric might provoke violence either by making it seem righteous or by disturbing people to the point of totally irrational behavior. The latter worry is not merely conjecture: One infamous group who feared that AI would end the world turned into a cult and was then connected to several murders (though none of the killings appeared to have anything to do with AI development).

“There is this kind of an apocalyptic mindset that people can get into,” Torres told me. “The stakes are enormous and literally couldn’t be higher. That sort of rhetoric is everywhere in Silicon Valley.” He never worried that anybody in Stop AI would resort to violence; he was always more freaked out by the rationalist crowd, who might use “longtermism” as a poor ethical justification for violence in the present (kill a few people now to prevent extinction later). But he did think that committing to an apocalyptic framing could be risky generally. “I have been worried about people in the AI-safety crowd resorting to violence,” he said. “Someone can have that mindset and commit themselves to nonviolence, but the mindset does incline people toward thinking, Well, maybe any measure might be justifiable.”

Ellen, the Stop AI adviser, shares Torres’s concern. Although he wasn’t present for what happened with Kirchner in November (Ellen lives in Hong Kong and has never met Kirchner in person, he told me), his sense from speaking frequently with him over the past two years was that Kirchner was under an enormous amount of pressure because of his feeling that the world was about to end. “Sam was panicked,” he said. “I think he felt disempowered and felt like he had to do something.” After Stop AI put out its statement about the alleged assault and the calls to police, Ellen wrote his own post asking people to “stop the ‘AGI may kill us by 2027’ shit please.”

Despite that request, he doesn’t think apocalyptic rhetoric is the sole cause of what happened. “I would add that I know a lot of other people who are concerned about a near-term extinction event in single-digit years who would never even consider acting in violent ways,” he said. And he has issues with the apocalyptic framing aside from the sort of muddy idea that it can lead people to violence. He worries, too, that it “puts the movement in a position to be ridiculed” if, for instance, the AI bubble bursts, development slows, and the apocalypse doesn’t arrive when the alarm-ringers said it would. They could be left standing there looking ridiculous, like a failed doomsday cult.

His other fear about what did or didn’t (or does or doesn’t) happen with Kirchner is that it will “be used to paint with a broad brush” about the AI-safety movement, that it will depict the participants as radicals and terrorists. He saw some conversation along those lines earlier last month, when a lawyer representing Stop AI jumped onstage to subpoena Sam Altman during a talk—one widely viewed post referred to the group as “dangerous” and “unhinged” in response to that incident. And in response to the news about Kirchner, there has been renewed chatter about how activists may be extremists in waiting. This is a tactic that powerful people often use in an attempt to discredit their critics: Peter Thiel has taken to arguing that those who speak out against AI are the real danger, rather than the technology itself.

In an interview last year, Kirchner said, “We are totally for nonviolence, and we never will turn violent.” In the same interview, he said he was willing to die for his cause. Both statements are the kind that sound direct but are hard to set store by—it’s impossible to prove whether he meant them and, if so, how he meant them. Hearing the latter statement, about Kirchner’s willingness to die, some saw a radical on some kind of deranged mission. Others saw a guy clumsily expressing sincere commitment. (Or maybe he was just being dramatic.)

Ellen said that older activists he’d talked with had interpreted it as well-meant but a red flag nonetheless. Generally, when you dedicate yourself to a cause, you don’t expect to die to win. You expect to spend years fighting, feeling like you’re losing, plodding along. The problem is that Kirchner, according to many people who know him, really believes that humanity doesn’t have that much time.

America Is Taking the Train

2025-11-15T06:00:00-05:00

You could almost mistake it for an ad. Last week, the far-right Georgia representative Marjorie Taylor Greene was on the Amtrak Crescent traveling from the nation’s capital to her home state, and she was enchanted. “The sweetest people run the train,” she posted on X, alongside a video of the autumnal landscape rushing by. “And the morning views of my north Georgia mountains made me smile and warmed my heart.”

As Greene said, she’d wound up on the train because of “flight delays and cancellations,” a result of the government shutdown. (Thousands of flights have been canceled over the past week, and delays have been common as unpaid air traffic controllers are overworked or walk off the job entirely.) Previously, Greene had been no particular fan of the train—in 2021, she voted against the Bipartisan Infrastructure Law that earmarked $66 billion to rehabilitate the country’s woefully out-of-date rail network—yet now she was embracing its charms.

[Read: What really happens after the shutdown ends]

Could it be that the chaos at American airports has created a small window of opportunity for the beleaguered rail system, which has so often been regarded as an embarrassing burden on the taxpayer? Is it possible that a form of transportation often associated with liberal wonks could, in the country’s hour of desperation, come through with a pleasant surprise for everyone? Is it Amtrak’s time to shine?

The shutdown is ending, but airports won’t be back to normal right away. Many people are turning to the train instead: Jason Abrams, a spokesperson for Amtrak, told me via email that the railroad “predicts record breaking 2025 Thanksgiving travel, with double digit growth in early bookings relative to last year.” In ordinary times, the Amtrak might appear laughable—it might seem ridiculous, for instance, that its trains have not been able to run between Albany and Boston or Albany and the Berkshires for the past several months because of a sinkhole situation affecting the tracks in Rensselaer County, New York. But these are not ordinary times, and a sinkhole is nothing compared with Newark airport in recent days. (Anyway, the sinkhole is almost fixed.)

[Read: There is no good way to travel anywhere in America]

Not everyone will be able to take a train instead of flying, obviously. Last year, 1.2 million people traveled by train for Thanksgiving, a pitiable number relative to the roughly 18 million who flew. Still, Jim Mathews, president and CEO of the nonprofit Rail Passengers Association, told me that he is “optimistic” about Amtrak’s potential to make a positive impression.

He suggested that an uptick in ridership this year could lead some to adopt Amtrak permanently and compared the post-shutdown moment, somewhat grimly, to the aftermath of September 11. Before 9/11, the most common way to travel between Washington, D.C., and New York City was via plane. Afterward, people were briefly afraid to fly, but flying also became inconvenient because of the added stress and time-sink of going through security. People tried the train and realized that it was easier and more comfortable; they could work the whole time, and they could go to a café car. In New York, they arrived in Midtown Manhattan instead of at an airport deep in Queens. And so they kept taking the train.

During the government shutdown, a lot of flight cancellations were on shorter routes. Even when that kind of flight isn’t canceled outright, a delay of something like two hours can tilt the math and make taking a train more logical. The Amtrak ride between Detroit and Chicago takes five hours, and New Orleans to Mobile, Alabama, takes four—each perhaps more appealing than getting to the airport at least an hour ahead of your flight, which may well be delayed anyway. Then there are the other, much-touted perks of trains: no TSA; no middle seat; you can bring 100 pounds of luggage for free. Like Representative Greene, some will sit down on a train in the coming weeks for the first time in a long time, look out the window, and remember that our country, despite its current acute dysfunction, is remarkably beautiful, a pleasure to see from a new angle.

Robert Paaswell, a civil-engineering professor at the City College of New York and a very expressive train enthusiast, told me that if he were running Amtrak, he would have full-page ads in major newspapers right now. In every paper, these ads would explain to people how to find the train schedule, how to buy a ticket, how to get to the station. “That’s a generational thing,” he said. People don’t even think of the train anymore. They don’t know where it is. (I’m not sure they know where to buy a newspaper either, though.)

Actually, Amtrak does market itself fairly successfully to young people as a more appealing and romantic alternative to driving and air travel, if not a more practical one. (Over the summer, for a social-media campaign, it made the interesting choice to depict an Amtrak locomotive smashing a mini bag of airplane pretzels, a tiny airplane window, an uncomfortable airplane seat, and an entire airplane as it flew over a city skyline.) Over the years, Amtrak has had several great marketing campaigns, including the “We’ve Been Working on the Railroad” ads from the 1970s, which Rick Harnish, executive director of the nonprofit High Speed Rail Alliance, referred to when we spoke this week. “They did a lot of advertising to get people to come back,” he said. “And then the service wasn’t good enough to keep them.”

[Read: A ‘death train’ is haunting South Florida]

To name a few issues, long-distance routes are prone to dramatic delays, and the cost of sleeper rooms is extreme; you can pay thousands of dollars to spend multiple days riding in a tight space that is not nearly as cool as anything you’ve seen in a classic movie. Amtrak train cars are in many cases decades old, and the railroad is currently short on equipment, which leads to cancellations—followed, sometimes, by seemingly random un-cancellations.

And service is sparse. Every year, I take the train home to Rochester, New York, for Thanksgiving, which is not an eccentric thing to do but is also not common. Only about 159,000 people got on or off a train in Rochester in all of 2024. So, logically, only four trains from New York City stop there every day (each carrying only a few hundred people). There’s a vicious cycle: Amtrak is underfunded and thus underused, which leads it to be further underfunded. (Why pay for a service that isn’t popular because it’s bad?) Now a significant percentage of travelers who check the Amtrak site out of curiosity will find that the train doesn’t go where they need it to, doesn’t go often enough, takes too long, or is too expensive.

When I looked this week at tickets for various routes, many of the trains for the days before Thanksgiving were sold out. Those that remain are a test of a person’s mettle—would you pay $500 to sit in an upright position for more than 20 hours to get from New York to Chicago? Would you sign up for a 10-hour-and-40-minute layover in the middle of your multiday trip from Raleigh to Syracuse, which on an ordinary day is a less-than-four-hour flight?

I suppose you wouldn’t, in regular times. This year, your choices are limited. The train may be slow, dingy, and stupid, but at least it stays on the ground.

A ‘Death Train’ Is Haunting South Florida

2025-10-22T07:00:00-04:00

Photographs and videos by Aleksey Kondratyev

Updated at 1:55 p.m. ET on October 22, 2025

This article was featured in the One Story to Read Today newsletter. Sign up for it here.

The Brightline is a beautiful train. Ultra-quiet and decorated with streaks of highlighter yellow, it carries passengers between Miami and Orlando, sometimes moving as fast as 125 miles per hour. It restores glamour to the humble railroad: During your ride, if you wish, you can order a half bottle of Veuve Clicquot for $59; the on-board bathrooms are large and clean enough to take a decent mirror selfie in. Condé Nast Traveler has called it “super chic.”

Privately owned and operated and transporting about 250,000 passengers a month, the Brightline is only the second high-speed train in the United States and the first outside the Northeast Corridor, where Amtrak operates the Acela. Its newness and sleekness make it a novelty in a country where trains are mostly old and ugly. Its existence shows that America can still build great things and that private industry can build them quickly and with style. If a beautiful high-speed train can work in Florida—whose former governor famously rejected more than $2 billion in federal funding for such a train—maybe it can work anywhere. But right now, something is very wrong.

What the Brightline is best known for is not that it reflects the gleam of the future but the fact that it keeps hitting people. According to Federal Railroad Administration data, the Brightline has been involved in at least 185 fatalities, 148 of which were believed not to be suicides, since it began operating, in December 2017. Last year, the train hit and killed 41 people—none of whom, as best as authorities could determine, was attempting to harm themselves. By comparison, the Long Island Rail Road, the busiest commuter line in the country, hit and killed six people last year while running 947 trains a day. Brightline was running 32.

In January 2023, the National Transportation Safety Board found that the Brightline’s accident rate per million miles operated from 2018 to 2021 was more than double that of the next-highest—43.8 for the Brightline and 18.4 for the Metra commuter train in Chicago. This summer, the Miami Herald and a Florida NPR station published an investigation showing that someone is killed by the train, on average, once every 13 days.

Floridians have started calling it the “Death Train” and maintain a sense of gallows humor about it, saying that it must be “fed” regularly to keep hurricanes away. Train attendants told me that Brightline engineers and conductors sometimes darkly joke about earning a “golden ticket”—which is when the train hits someone at the right time so that the three paid days off a worker gets for emotional distress are rolled into a weekend that takes up most of the week.

Brightline argues that the “Death Train” moniker is unfair for many reasons. One is the notorious difficulty of determining whether a death on a train track was a suicide. The company says the true rate of suicides on its Florida route is higher than government agencies report because of the variability in how local law-enforcement agencies and medical examiners make their determinations. Although Brightline no longer insists, as it has in the past, that the majority of the deaths are the result of suicides or drugs, it still takes care to frame the issue as a matter of personal responsibility. None of the deaths on Brightline tracks has been the result of equipment failure or operator error, Ashley Blasewitz, Brightline’s director of media relations, wrote to me in an email. “All have been the result of illegal, deliberate and oftentimes reckless behavior by people putting themselves in harm’s way.”

Federal agencies have investigated the Brightline incidents and produced no firm conclusions about why they have happened so often. The company, sometimes called “Frightline” on the local news, has not been found responsible for any of the deaths. How could it be responsible for people driving around lowered gates or walking into the clearly delineated path of a train? Yet there must be some explanation for the unusual number of fatalities.

Brightline’s parent company aspires to create additional train routes all over the country. It has been embraced by pro-transit wonks and former President Joe Biden’s train-nerd transportation secretary, Pete Buttigieg, as well as by tech-world influencers and members of the Trump administration. In a February press release announcing that it would investigate a federally funded California high-speed-rail project that has become a decade-plus boondoggle, Donald Trump’s Department of Transportation praised Brightline by comparison, citing its “impressive work” on Brightline West, the company’s second route. Still under construction, Brightline West will connect Las Vegas to the Los Angeles suburbs with a train that can go up to 200 miles per hour.

If Brightline really is the future of rail in the United States, the most important question is obvious: Why are so many people dying?

Aleksey Kondratyev for The Atlantic

Ultra-quiet and decorated with streaks of highlighter yellow, the Brightline carries passengers between Miami and Orlando, sometimes moving as fast as 125 miles per hour.

A popular theory of the Brightline deaths, which you’ll see in comments underneath viral videos of the train plowing into cars, is that there is something wrong with people who live in Florida. Specifically, these comments invoke the concept of the “Florida Man”—a long-standing meme that suggests the state is, in essence, full of morons.

Jim Kovalsky, the president of a nonprofit called the Florida East Coast Railway Society, appeared exasperated in a local-TV interview last year. “If you don’t put yourself between those two steel rails, you’re not going to get hit by a train,” he said. When I spoke with him earlier this year, he was even more direct. “I think the concept of Florida Man is real,” he said. “Unfortunately, we are dealing with a lot of people that don’t understand self-preservation.”

But if the people of Florida were uniquely stupid in a way that made them more susceptible to being hit by trains, you would expect them to be hit uncommonly often by all trains. This is not the case. Amtrak serves fewer passengers than Brightline, but operates through many of the same urban areas as well as some additional ones, and it reported six total fatalities in the state in 2024, compared with Brightline’s 41. The NTSB’s 2023 report found that Brightline’s accident rate per million miles was more than eight times that of SunRail, another commuter train that operates around Orlando. Brightline has challenged the usefulness of this statistic, noting that it doesn’t account for the amount of daily traffic around and on the tracks, but that is sort of the point.

The Brightline runs on the route of the original Florida East Coast Railway, which was built in the late 1800s by Henry Flagler, a Standard Oil tycoon. Flagler is popularly credited with “inventing” modern Florida: His railroad allowed for the development of swampland into a series of luxury resorts dotting the coast. Everything grew up around this track—it’s the vein running through all of the oldest cities and most densely populated areas of South Florida.

Passenger trains stopped running on this line in the late 1960s, leaving it to slower freight trains that ran less frequently. When Brightline’s parent company, Florida East Coast Industries, was taken over by the private-equity firm Fortress Investment Group, it built a second track so that passenger trains and freight trains could efficiently share the space. (Then it sold the freight rights to a Mexican conglomerate for $2.1 billion.) Since 2017, far more trains than ever before have run through these areas, and faster, in both directions at the same time.

Railway signals along the Brightline tracks in Hollywood, Florida

The Brightline heads south to Miami from Fort Lauderdale.

Aleksey Kondratyev for The Atlantic

As a result, once-familiar environments have been transformed. Take, for example, the story of Joann DePina, a 49-year-old mother of two who was killed by a Brightline train in January. DePina was walking over the tracks that cut through her neighborhood, but she was doing so on a well-worn footpath. She was technically trespassing, but there weren’t any fences or no trespassing signs, and it was a logical thing to do. DePina rented a room in a sober-living house on one side of the tracks and was crossing to get to a group meeting on the other side. She had been in recovery since 2017 and was saving money to move into her own apartment.

I walked along the tracks with her aunt Maria Furtado in May. Furtado showed me the footpath, next to the white cross she’d put up in her niece’s memory. In person, it was clear why people would walk there: The tracks split the neighborhood in half, with tightly packed houses on one side and a row of businesses on the other. To get around the tracks legally would require walking down to an intersection to cross, then walking back, adding at least 10 minutes. Taking a shortcut over the tracks looks easy enough, and it was probably easy to do so safely during the decades when freight trains were the only traffic. Hence the worn path.

“I worry about these people all the time,” Furtado told me, gesturing at a house whose yard ended less than 50 feet from the tracks. On a previous visit, she’d seen a young boy chasing after a cat as it walked on the tracks. As we talked, Furtado pointed behind me. I turned around and saw a Brightline train coming toward us—only a few seconds away, at most. The train whipped past—it’s powered by quiet diesel-electric locomotives and goes 79 miles per hour through that part of its route. It was easy to put myself in DePina’s place. She was walking at night, and she didn’t hear or see anything coming. Her timing was horrible.

After DePina’s death, Furtado attempted to contact Brightline but never heard back. She also contacted Governor Ron DeSantis, who forwarded her letter to the Florida Department of Transportation, which gave her a polite but vague response about its commitment to safety. (During the course of my conversations with Brightline about its record, Blasewitz provided a list of safety improvements that had been made to the tracks both before and after the train started operating, which cost nearly $500 million. “Brightline is one of the safest forms of transportation in the state of Florida, moving millions of people out of their cars and off dangerous roadways,” she wrote by email.)

Furtado told me she wasn’t sure what other options were left to her. “I don’t know who to blame,” she said. In her opinion, someone should have to put up a fence along parts of the tracks that cut through neighborhoods—whether that’s the city or the state or Brightline, she doesn’t much care. Being from Massachusetts and having some familiarity with northern commuter trains, she also liked the idea of the tracks being elevated, even a little bit, to deter people from walking over them. “She wasn’t going to hike a mountain or climb over a fence to get across,” she said of her niece.

DePina’s story is elucidating, but it’s only one incident. The NTSB has been conducting a series of investigations into Brightline accidents to search for patterns and will eventually publish a summary analysis of those findings. But so far, only a handful of reports have been published, and they offer few clear takeaways.

For instance, last year, investigators looked into a pair of fatal accidents that had happened two days apart at the same intersection in Melbourne, Florida, a small, coastal city 70 miles southeast of Orlando. Both involved drivers going around safety gates. The details of the first crash were especially odd. The crossing’s gates and all of its other safety devices had been working perfectly. Neither the engineer nor the conductor of the train had done anything wrong, while the driver of the car did at least two obvious things wrong. The first was that he had driven around a stopped car and then the lowered gate even as a woman in the back seat of his car yelled at him not to. The second, a toxicology report showed, was that he had been on bath salts.

On the one hand, the issue here was obvious: Florida Man. On the other, the NTSB’s report contains information that suggests a dangerous environment, regardless of one’s drug intake. It noted that Brightline service had dramatically increased train traffic through Melbourne in recent years. The double-tracking of the line at this location had been completed in June 2023; previously, 14 freight trains passed through each day, and now there were 14 freight trains plus 32 higher-speed passenger trains. Before the two back-to-back Brightline incidents, there had been only three crashes since 1975.

As part of that investigation, NTSB staffers rode the Brightline one Sunday from Orlando to West Palm Beach and back. They found all crossing gates and warning lights to be functioning perfectly and the train crews to be professional and alert. Yet the train they rode had to make an emergency stop to avoid hitting a pedestrian in Melbourne. Then it nearly hit a bicyclist, also in Melbourne. “While talking with the engineer,” the investigators noted in their write-up, “he stated that he had been involved in seven incidents while working for Brightline involving striking trespassers or vehicle strikes.”

Aleksey Kondratyev for The Atlantic

The Brightline’s track intersects flatly, or “at grade,” with the roads on much of its route, cutting through the landscape at strange angles and in unexpected places.

Many train tracks are elevated to cross above roadways. Others are sunken down to cross beneath them. But the Brightline’s track intersects flatly, or “at grade,” with the roads on much of its route, including the part that runs through central Miami.

Many states have undertaken grade-crossing-elimination projects over the past half century because they make train routes dramatically safer. On the Amtrak route between Washington, D.C., and New York City, the highest-trafficked stretch of train track in the country, there are no grade crossings. The last one was eliminated in the 1980s.

There are 331 grade crossings along the Brightline route in South Florida. James Hopkins, a former Brightline conductor, cited this when explaining to me why he no longer works for the company. He mostly enjoyed his time at Brightline, he said—the company was a good employer—but he didn’t want to work on that route anymore in large part because of how often the train would hit people. At his previous job operating a freight train in the 200-mile stretch between Pittsburgh and Harrisburg, he said there were 40 to 50 grade crossings. In the 65 miles between West Palm Beach and Miami, there are 174. “It’s just real busy,” he told me. “The fatalities—this was just something I didn’t want to continue doing.”

When I visited the West Palm Beach area to look at the crossings and roads in person, I drove over the tracks dozens of times. They cut through the landscape at strange angles and in unexpected places—behind the downtown courthouse, alongside a Little League field in Delray Beach. People have been struck and killed by Brightline trains at both of these locations.

During my trip, I met with Eric Dumbaugh, a professor of urban and regional planning at Florida Atlantic University who has lived in the area for most of his life. “Brightline is unique nationally,” he said. “It’s operating right through the urban fabric.” Just by leaving their houses, people encounter it, whether they want to or not, and they sometimes have to react quickly, in a life-and-death situation, to a system they don’t intuitively understand. “This is why we see the issues that we have,” he said.

To visualize this, we drove to a grade crossing in Delray Beach, where an elderly couple had been hit and killed by a Brightline train in 2023. The NTSB investigated this accident, perhaps because it was so confounding. The couple had been driving down a road adjacent to the tracks just after 8 p.m. It was winter, so it was dark. The husband was behind the wheel—he had a green light, so he turned right, at which point there was only a short bit of roadway before the couple found themselves on top of the southbound track. They either didn’t register or didn’t have time to react to the gate’s warning lights and bells. Their timing, like Joann DePina’s, was horrible, and the gates came down while the car was in the crossing.

The couple apparently watched, unpanicked, as a northbound freight train approached from one direction. A witness told the NTSB that she saw the wife get out of the car, look around, then go over to the driver’s-side window to say something to her husband before getting back in the car. The woman had seemed calm. The best guess that Dumbaugh came to—the same as the NTSB’s—was that the wife had examined the car’s position and seen that it was clear of the track on which the freight train was approaching. She couldn’t see a train coming on the other track, from the other direction. The couple must have decided to wait for the freight train to pass. They turned off the engine of their car, as well as their headlights.

Just as the freight train passed, the Brightline came on the other track. It hit the front of the car and sent it spinning off the road, flipping onto its side. The wife was thrown out of the vehicle, while the husband was stuck inside. Both were dead at the scene. The witness pulled her own car onto the grass and sat for 10 or 15 minutes, shaking, she told the NTSB. She hadn’t seen the Brightline coming either.

Aleksey Kondratyev for The Atlantic

The Brightline heads south through Third Street and Dixie Highway in Pompano Beach.

Dumbaugh explored the intersection, pointing at various elements. Signage on the adjacent road made it clear that a train passes nearby, but didn’t tell drivers to be wary of turning into its path. “There was nothing on the approach that warned people a right turn would be an issue,” he said. From the NTSB report, we knew that the freight train had radioed the Brightline once it saw the car on the tracks and that the Brightline engineer had implemented the train’s emergency brakes, but there hadn’t been nearly enough time to stop. Again, it was a story without one easily identified insight. The Federal Railroad Administration regulates the operation of the gate. The road the couple had turned off was a state highway. The intersecting street was the responsibility of the city, but the traffic signals were the responsibility of the county.

Brightline says that it is an advocate for closing certain crossings on its route, but that this rarely happens “without local support.” Because of all the elements at any intersection, the process of closing even one crossing can be convoluted and expensive. Sealing off the entire Brightline route or elevating the entire track would simply not be economically feasible for a private company.

Still, over a period of months, I spoke with several experts who had different opinions on many of the technical details but who all agreed that there’s no real mystery behind the Brightline deaths. “Fast trains and grade crossings are always a deadly combination,” the historian Richard White, whose 2011 book about American railroads was a Pulitzer Prize finalist, told me. He put it the most succinctly, but I did not talk with anybody who disagreed with that conclusion.

While I was in Florida, I hoped to hear directly from Brightline executives. The company was co-hosting a conference called the Railway Interior Innovation Summit, in Orlando. So, of course, I took a Brightline train to get there.

The train ride was unlike any I’ve taken in my life. The Brightline’s passenger cars are softly lit with pretty blue LEDs along the ceiling, and the roomy seats are upholstered with soft white leather. There is ample legroom and nothing is broken. The elegant new stations have cocktail bars named Mary Mary, apparently in reference to Henry Flagler’s first and third wives, who had the same first name. The stations also have gift shops, where you can buy attractive Brightline merchandise—pink ball caps, soft sweatshirts, a candle matching the custom scent that is piped into the terminals.

I bought a “Premium” (first-class) ticket from West Palm Beach for $99, which came with a steak sandwich on a brioche bun for dinner, a passion-fruit tartlet for dessert, a dark-chocolate Lindt truffle for a second dessert, and a glass of cuvée from the complimentary-drink menu. The Brightline is the first train line to offer basically flawless Wi-Fi provided by Elon Musk’s Starlink, which is why I got to see Brett Baty put the Mets up over the Red Sox while hurtling up the coast at the end of a long day. The ride was smooth and quiet and we were exactly on time. We made it to Orlando without incident in two hours and 12 minutes—more than an hour faster than the typical Amtrak on this route, and much less stressful than driving a car.

Aleksey Kondratyev for The Atlantic

The Brightline’s passenger cars are lit with blue LEDs along the ceiling, and the seats are upholstered with white leather.

This probably would not have seemed remarkable to the rail summit’s many European attendees, whose countries already have high-functioning train systems. Many of these people were in the United States for the first time—meaning that their first experience of our wonderful and interesting country was three days in Orlando. At a networking event, I entered a cluster of conversation just as a Swiss man was explaining that American train stations are surrounded by “car parks,” which he found shocking, because most people in Switzerland ride their bikes to the train stations. (Switzerland is about half the size of Maine, by the way.)

A packed conference room listened to a panel on train start-ups, including one called Dreamstar Lines, which intends to begin operating a “hotel on rails”—a luxury overnight train between San Francisco and Los Angeles—before the 2028 Summer Olympics (in a mock-up, it had an on-board spa). Various companies showed concepts for spectacular and futuristic train cars, but Brightline was the center of attention. Its executives gave the most well-attended talks, got the biggest laughs. Everybody agreed that Brightline’s trains were impressive and that its proposals were exciting.

On the third day of the summit, participants were led on a tour of Brightline’s Orlando maintenance facility by Tom Rutkowski, the company’s vice president and chief mechanical officer, a former general superintendent for New Jersey Transit, and a charming, brassy host. When we all boarded one of the trains to look around, Rutkowski encouraged us to sit down and feel the leather, which he said was the same that is used in Bentleys. “If you’ve never sat in a Bentley, this is as close as you’re going to get,” he told us.

Aleksey Kondratyev for The Atlantic

The interior of a Brightline train as it heads south to Miami from Fort Lauderdale

After the formal tour, the group was offered complimentary wraps for lunch in a meeting room. When I walked in, Rutkowski was sitting on top of a table, holding court in front of a small group of men who were standing around asking him friendly questions about Brightline’s business. “We are poor,” he told them. “I’m lucky I can make payroll.” He said it not as if the company were desperate but as if it were scrappy. He added, “There’s no government money coming to bail us out.” (Rutkowski later denied making these comments, and called them “nonsense.”)

Some additional context is needed here. The claim that there is no public money coming to or already in use by Brightline is not exactly true: The Florida line was built, in part, with $2.2 billion of tax-exempt bonds. If Brightline were, for some reason, to go bankrupt, it might behoove either Florida or the federal government to bail it out and take over operation of the line, rather than leaving everything to rust and the hundreds of thousands of people who use the train to go back to their cars.

The bonds underlying Brightline have been downgraded multiple times this year because of slower-than-expected ridership growth and higher-than-expected costs. In July, the company announced its intention to defer interest payments on $1.2 billion of debt, and NPR reported that Brightline had been looking for outside investors for months with little success. Blasewitz, the media-relations director, told me that Brightline is still confident in its year-over-year growth and that it intends to establish itself as an “integral” part of Florida’s transportation system, though she declined to comment on when the company expects to become profitable.

In addition to the safety conversation, then, there is a conversation to be had about whether Brightline is even a private solution to a public problem at all. The new line in California, Brightline West, will be privately operated, but is being built with billions in federal grants. To the extent that I heard any muttering at the summit that was less than complimentary to Brightline, it was on this point.

At one summit event, I chatted briefly with Jim Mathews, the president and CEO of the nonprofit Rail Passengers Association, who thought the Brightline project was interesting and in some ways laudable. Still, he said that Brightline’s Florida strategy was not repeatable. It had been a quirk of history that its parent company owned the right-of-way on those old railroad tracks, and it would not be in that situation again anywhere else in the country. Plus, Brightline lost more than $500 million in 2024 while serving only six stops, he pointed out. Amtrak, often regarded as an albatross around taxpayers’ necks, lost more—about $705 million—but serves more than 500 stops, including many that a private enterprise would never bother with and that a public one is obligated to serve.

“The idea of scaling on a private level is just complete insanity,” Mathews told me when we talked again after the summit. “Brightline got 3 billion federal dollars to bring along Brightline West—which is great; I don’t oppose that. The more the merrier—the more service we have, the better it is,” he said. “But let’s not pretend this is the kind of capital investment that private industry can do by itself. They can’t.”

Aleksey Kondratyev for The Atlantic

“Fast trains and grade crossings are always a deadly combination,” says the historian Richard White.

If the most obvious question to ask about Brightline is Why are so many people dying on this one stretch of train track in Florida?, the second-most-obvious is Who can fix it?

It seemed to me that the problem in Florida was being treated as unsolvable, as though this is somehow just the way it is. The Federal Railroad Administration, for instance, doesn’t believe that Brightline is at fault for the frequent accidents. James Payne, the FRA’s staff director of grade crossing and trespasser outreach, told me frankly that South Florida is a mess. “It keeps me up at night,” he said. But in his opinion, Brightline is doing about as much as it possibly can to improve grade crossings and encourage safety, given the constraints of its business and the existing infrastructure.

I talked with Jim Mathews about the situation at some length, hoping for clarity. Mathews didn’t have a perfect explanation either. He thought Brightline had been arrogant and callous, but he also thought the real issue was bigger. Americans are okay with tens of billions of tax dollars funding highways and airports overseen by powerful regulatory agencies. But we don’t want to spend the same way on trains, even though we want trains to be built. “We love private industry because it doesn’t cost us money, but we point fingers at private industry when it kills people,” he said. “That’s why we have governments—they protect people; they step in where markets fail.” Or they should.

Just after the Miami Herald’s July story on Brightline deaths came out, Transportation Secretary Sean Duffy remarked that there had been “way too many deaths” in Florida, and that something should be done. Shortly after that, federal grants worth more than $42 million, which were awarded to Brightline between 2022 and 2024 but had not been disbursed, were finally ushered along by Duffy. Those funds will be used to make some safety improvements, including fencing along parts of the Brightline’s route and various interventions to deter people from driving around lowered gates. At the same time, the proposed 2026 Department of Transportation budget that was advanced by Congress over the summer includes no funding at all for the Federal Railroad Administration’s Crossing Elimination Grant Program, which is the primary means by which local governments all over the country have funded grade-crossing-removal projects.

In May, when I rode a Brightline train out of Miami, looking through the window at a ludicrously flat landscape, I thought about the future. The train hurtled through towns that were arranged on either side, going so fast while so close to houses, restaurants, parks, and people that I was startled again each time I looked out the window. This is not what it will be like when people ride Brightline West. That train will go through the desert and run mostly within an existing highway median. It won’t have the same pitfalls as this first experiment, for which people are dying and that’s just the cost of something new.

Later that evening, I scrolled on my phone and came across an Instagram post about another Brightline accident, with a caption describing the person who had reportedly been hit as a “track snack.” People in the comments responded jubilantly, praising the train for chowing down on another soul. The beast was getting stronger, the commenters said with satisfaction. “As always sorry if this was your family member,” the account runner wrote dutifully in the replies.

This article originally misidentified a former Brightline conductor and misstated the amount of interest the company owed on its debt.

The End of the Old Instagram

2025-10-17T16:03:00-04:00

Updated at 10:14 a.m. ET on October 21, 2025

Thirty years ago, parents everywhere were compelled to weigh the pros and cons of allowing their kids to see Titanic. At the time, it was the biggest movie ever made, a historical epic (potentially educational) about mass death (possibly traumatizing) with a romantic plotline that was maybe too exciting (you know what I mean!). It was rated PG-13—a guideline that recommended caution but ultimately ruled the movie to be appropriate for millions of teenagers—resulting in a fortune for its creators and the subsequent blessings of Leonardo DiCaprio’s career.

Instagram is now adopting the same label for a teen-safety feature, but the possible outcomes are less discrete and obvious. Meta announced earlier this week that all Instagram users under the age of 18 will be automatically placed in what it’s calling a PG-13 version of the app, where only content that might appear in a PG-13 movie will, ideally, be visible. Thousands of parents were surveyed to help create the new guidelines. “We hope this update reassures parents that we’re working to show teens safe, age-appropriate content on Instagram by default,” the company wrote in a news post.

[Read: The biggest change to Instagram in years]

This is an update to an existing Teen Accounts feature, which already sought to limit exposure to graphic violent and sexual content, as well as to posts promoting cosmetic procedures and eating disorders, alcohol and tobacco sales, and other things that parents frequently worry about their kids seeing online. Although the PG-13 rating would seem to give a lot of leeway, it’s actually more restrictive than the system that was in place: It expands the internal list of worrisome content. Now, according to the update, posts about “certain risky stunts” may also be hidden, for example, while posts containing “strong language” will be removed from teens’ recommendations. Accounts that regularly share inappropriate things will be hidden from users under 18.

Whereas the old version of Teen Accounts applied wide content restrictions to users ages 13 to 16 by default, but allowed more flexibility to 17-year-olds, the PG-13 setting now restricts them as well—strange, given that they can see an R-rated movie, to extend the movie metaphor. (Parents who are feeling more or less permissive are able to toggle from the PG-13 default to a more heavily filtered Limited Content setting or a relatively lenient More Content setting.)

Liza Crenshaw, a spokesperson on the youth-and-well-being team at Meta, told me that the company has consistently heard from parents that they are concerned about what their kids are seeing online and that they are generally unclear on what is and isn’t allowed by platform policies. I can understand the tricky position Instagram is in, but the update feels oddly simplistic, unlikely to satisfy people who are reasonably concerned about an array of highly complex problems. After at least half a decade of acute concern about the way that platforms such as Instagram may affect young people, as well as intense debate about how best to keep kids safe online, Meta has arrived at a label that was invented in the 1980s because parents were upset by movies such as Gremlins.

Content-moderation experts I spoke with this week were baffled. “Why would Meta choose a system that doesn’t graft neatly onto social media, but was rather designed in another era?” Shauna Pomerantz, a professor of child and youth studies at Brock University who studies social media, told me when I sent her the press release.

When I asked, Crenshaw suggested that PG-13 is a simple and legible construct with which parents are already familiar. “They wanted an easier way to understand the guidelines,” she told me. Making PG-13 the baseline—the place where all users ages 13 to 18 start out—was clarifying for them. It gave them a quick mental frame of reference because they know, basically, what is and isn’t allowed in a PG-13 movie from years of exposure to that rubric. James Grimmelmann, a professor of digital and information law at Cornell Law School who teaches courses on content moderation, called it “a Gen X–nostalgia move.”

Additionally, the PG-13 rating seems to imply that something fundamental has changed about Instagram—that Meta now views the platform more like a movie theater, where audiences passively witness content, than a social network, where individuals engage in ongoing meaning-making together. Even if there’s some truth to the idea (surveys from Gallup and Pew have shown that users are more likely to consume content than post it themselves), Instagram posts are inevitably part of larger conversations that evolve moment to moment.

Grimmelmann noted that the new system is a blunt intervention. “It means that Meta is backing away from the most important piece of content moderation: context sensitivity.” He suspected that plenty of inoffensive content would be taken down accidentally, because such a sweeping moderation effort “won’t have the nuance to appreciate differences between showing, glorifying, discussing, and criticizing.” The PG-13 system will certainly steer teenagers away from content that could be truly harmful to them—a good thing!—but it may also curtail information and creative expression.

For example, you can imagine inadvertent filtering of conversations about violence, sexuality, or substance use that may actually be appropriate for a 17-year-old who is about to enter college or the workforce. The PG-13 rating raises obvious questions about, say, disturbing news footage that might be discussed in a social-studies class but filtered by Instagram. And does it really make sense that a teenager would be unable to follow the pop star Charli XCX, because she posts photos of herself smoking cigarettes, or an account aggregating stills from French films, because the people in them are sometimes nude?

[Read: AI’s emerging teen-health crisis]

All of these problems are only becoming more complicated, and weirder, in the age of artificial intelligence. The announcement from Meta also noted briefly that the PG-13 restrictions will apply to its AI-chatbot and AI-character features: “AIs should not give age-inappropriate responses that would feel out of place in a PG-13 movie.” Stupid and ugly AI-generated content, such as this video my colleague sent me of a man barfing up a stream of whole hot dogs, is not mentioned. (Parents will also be able to restrict teens’ access to AI tools.)

Many of the efforts to help kids navigate social media are sold with the throwback language of abstinence and purity. For instance, the Wait Until 8th movement asks parents to sign a pledge saying they won’t give their kids smartphones until eighth grade, and the Appstinence organization, a college-student-led group, encourages young people to delete social-media apps. The social psychologist and author Jonathan Haidt’s best-selling book The Anxious Generation, which has become a bible of sorts for many parent groups, draws an explicit dichotomy between the good childhoods of previous generations—when kids rode bikes to one another’s houses, made tire swings, and entertained themselves—and the worse childhoods of today. One of Haidt’s oft-repeated ideas is that modern kids are dramatically over-supervised offline and dramatically under-supervised online. He argues that kids should be empowered to move around the offline world alone more often, but that social media should not be available to kids under 16 at all.

When I spoke with Crenshaw, I had this cultural backdrop in mind. I mentioned that a lot of the broad content-moderation efforts of the recent past ended up stifling conversation about many of the things for which teenagers most value social media. They go online to express themselves creatively, explore their identities, and hear from other people about their experiences—maybe crassly at times, or in ways that are uncomfortable to adults, but that isn’t necessarily harmful. “It’s a really challenging one and something we always have to balance when it comes to content,” Crenshaw told me. “Where we are right now is we really want to err on the side of caution when it comes to teens’ experiences.”

Listening to parents is a welcome departure from Instagram’s past, when moderation decisions were made entirely behind closed doors. The new PG-13 system will regularly survey parents to ask whether individual posts seem appropriate or inappropriate for teens—which some may find annoying and needlessly time-consuming, and others may welcome as evidence of a new respect and conscientiousness.

Still, the anachronistic label may ultimately be a false comfort. People may long for the days when questions were as simple as “Should I let my daughter see Titanic?” But that’s not the world we live in.

This article originally misstated how parental surveys were used to help update Meta’s content policies.

The Trumpian Fantasy of WhiteHouse.gov

2025-10-09T13:27:49-04:00

Last week, Donald Trump’s White House anticipated the impending government shutdown like an album release, placing a massive countdown clock at the top of WhiteHouse.gov. “Democrat Shutdown Is Imminent,” read the online home of the People’s House, on a black background. Now that the shutdown has happened, a clock is counting upward: “Democrats Have Shut Down the Government,” it says, with numbers climbing to mark the seconds, minutes, hours, and days that have elapsed.

This is an unusual use of the White House website. Though WhiteHouse.gov has always been a place to showcase the administration’s agenda, it has mostly looked like the website of a mid-size high school. During the Clinton administration, it had the goofy GeoCities look of the day (American-flag GIFs); by the start of George W. Bush’s presidency, it had transitioned into a bland informational page rendered in blue, white, and gray, clotted with text. (“President Bush Participates in Signing Ceremony With NATO Secretary General De Hoop Scheffer for NATO Accession Protocols for Albania and Croatia,” for example.) It stayed that way, with minor tweaks, throughout the Obama administration, and it was as dry as ever during Trump’s first term too. Even as Trump was inciting an insurrection against the United States government, his team did not use the White House website to promote that goal. On January 6, 2021, the homepage still showed information about the new COVID-19 vaccines.

But when Trump returned to office in January 2025, his transition team had a redesign ready to go. The first day, the website was transformed. Visitors saw an auto-playing trailer with an action-movie score—helicopter, jets, eagle, salute, thumbs-up, then a new White House logo in which said house was mostly black. After the video came a landing page with a photo of Trump and the message “AMERICA IS BACK” written in a new, spindly serif font on a dark navy background. Unmistakably, the design evokes the concept of “dark mode,” the default app setting for guys who take themselves very seriously and who relish the idea that they may be edgy and cool. (A friend of mine used to react to people putting their phones in dark mode by saying “Okay, Batman.”) By the way, the site is no longer available in Spanish.

[Read: We’re all in ‘dark mode’ now]

Americans don’t need the White House website to explain to them the attitude of this administration—Trump’s actions and the consequences of them are plain to see. Yet the White House website is a record of an era: Looking back at the Bush years, I was struck by the plainness of the design, but also the gentle and classic expressions of patriotism that were about as jarring as an American-flag postage stamp. If WhiteHouse.gov is a chapter in the story of the second Trump administration, what is it saying?

Not a design expert myself, I asked Pamela Lee, a professor of modern and contemporary art at Yale, to take a look at the site. I told her I thought the dramatic darkening of the page scanned to me as creepy and menacing, but she called this a matter of perspective. “You read it as spooky,” she said. “Some folks might come to it and think it represents something serious, somber, and masculine.” (Appropriate, maybe, for a return to power.)

The same “dark mode” font treatment and color scheme have been used on the White House social-media pages since the first days of the new administration, marking another departure from the previous anodyne style. As my colleague Charlie Warzel wrote in March, on X, the White House is now a troll account, borrowing its snide visual language and tone from some of the internet’s most cynical spaces and deploying this style to mock and dehumanize people.

These updates are apparently part of a larger project. In August, Trump announced the creation of a National Design Studio led by an Airbnb co-founder and Tesla board member, Joe Gebbia, one of the DOGE figures who was seen as a successor to Elon Musk after Musk’s departure from Washington. (One of the National Design Studio’s first projects was the website for the Trump Gold Card, a U.S. visa that will be granted only to those who can “make a gift of $1 million” to “substantially benefit” the United States.)

This new team reportedly replaces a group of United States Digital Service and General Services Administration employees, many of whom resigned or were fired during the DOGE cuts earlier this year. It is tasked with modernizing the government’s digital services, but it also promises to beautify them. A launch page for the National Design Studio specifically names the Apple Store as a north star. (The White House initially responded to my request for an interview with the new team, but didn’t respond to subsequent attempts to schedule one.)

A week after announcing the design studio, Trump signed an executive order titled “Making Federal Architecture Beautiful Again,” which states that classical styles emulating ancient Greece and Rome are the new “default” for government buildings. This sounds like a bit of a mishmash, but I can kind of see the vision. It’s familiar as one that has been popular in Silicon Valley for years, where a survey might find that the most beautiful things ever created are Apple devices and the Roman empire.

This hybrid look is shared by many “network state” projects that have emerged in recent years. Those projects, which boast funding from the likes of Sam Altman, Marc Andreessen, and Peter Thiel, promise total freedom for people who regard themselves as overly constrained by our current democracy. They tend to combine elements of sleek, modern design with images and references drawn from the distant past, when men were great, spears were shiny, and buildings were intimidating. They like the look of Roman- and Greek-sculpture busts, for instance, but Photoshopped with gradient overlays and sci-fi elements. Another tech-world project called More Monuments is currently working on building a 500-foot-tall statue of George Washington in a classical style but made of stainless steel, which they are funding in part with a crypto token called GEORGE; they plan to call it The Colossus of George.

Trump’s personal taste is all over the place. He leans more toward the gilded, his own interior-design preference more toward Versailles. But his chosen architect for the gigantic new White House ballroom is a member of the National Civic Art Society, a nonprofit whose goal is promoting classical architecture, and his selection of Gebbia, who went to the Rhode Island School of Design and cites the Bauhaus movement as inspiration, suggests that he is on board with the Apple-meets-Rome combination.

When I spoke with Toby Norris, an art-history professor at Assumption University who contributed to the recent Routledge book Interrogating the Visual Culture of Trumpism, he said he didn’t think that Trump had a coherent aesthetic vision. Instead, he sees “a kind of patchwork of all these things that different people who have influence on him have been pushing.” The executive order on architecture, for instance, was reportedly “spearheaded” in 2020 by the National Civic Art Society. Trump issued a version of it at the end of his first term but it was invalidated by the Biden administration almost immediately.

When Trump presents the idea of a return to the classical, it’s in a populist tone. Both the 2020 and 2025 orders argue that people dislike the Brutalist government buildings of the second half of the 20th century, and that a revival of classical architecture would be a way of giving people what they want. Critics have countered that classical architecture has taken on a more authoritarian reputation over time. It’s the architecture of ancient Athens, the birthplace of democracy. “But it’s also the architecture of the Roman empire,” Norris said. The later classical architecture of Rome was on a grander scale—more imperial and assertive and over-the-top, he told me. “And then people point out that’s exactly what Hitler liked,” he added brightly.

At the end of the day, the “dark mode” online aesthetic paired with the offline return to a fantasy of the awe-inspiring past is not much more than a vibe—a porridge of references to power and control. When I spoke with Lee, she noted that the right has recently been reaching into the “grab bag” of history and looking for “moments that represented either the golden ages of this or that or kind of cusp moments.” And the gloomy website I pointed to seemed, to her, to represent a darkness before a dawn, if ham-handedly.

Whatever the intention, it would probably be easy enough to sell these ideas to Trump simply by calling them beautiful. “Trump uses the word beauty all the time,” Norris observed. “It’s obviously a sort of talisman for him, this word beauty.” People can disagree about what’s beautiful, of course. In her 1999 classic, On Beauty and Being Just, Elaine Scarry argued that spontaneous glimpses of beauty are what inspire in ordinary people the pursuit of truth and justice. I guess from another perspective, it could just mean “winning.”

AI’s Emerging Teen-Health Crisis

2025-09-18T17:59:55-04:00

On Tuesday afternoon, three parents sat in a row before the Senate Judiciary Subcommittee on Crime and Counterterrorism. Two of them had each recently lost a child to suicide; the third has a teenage son who, after cutting his arm in front of her and biting her, is undergoing residential treatment. All three blame generative AI for what has happened to their children.

They had come to testify on what appears to be an emerging health crisis in teens’ interactions with AI chatbots. “What began as a homework helper gradually turned itself into a confidant and then a suicide coach,” said Matthew Raine, whose 16-year-old son hanged himself after ChatGPT instructed him on how to set up the noose, according to his lawsuit against OpenAI. This summer, he and his wife sued OpenAI for wrongful death. (OpenAI has said that the firm is “deeply saddened by Mr. Raine’s passing” and that although ChatGPT includes a number of safeguards, they “can sometimes become less reliable in long interactions.”) The nation needs to hear about “what these chatbots are engaged in, about the harms that are being inflicted upon our children,” Senator Josh Hawley said in his opening remarks.

Even as OpenAI and its rivals promise that generative AI will reshape the world, the technology is replicating old problems, albeit with a new twist. AI models not only have the capacity to expose users to disturbing material—about dark or controversial subjects found in their training data, for example; they also produce perspectives on that material themselves. Chatbots can be persuasive, have a tendency to agree with users, and may offer guidance and companionship to kids who would ideally find support from peers or adults. Common Sense Media, a nonprofit that advocates for child safety online, has found that a number of AI chatbots and companions can be prompted to encourage self-mutilation and disordered eating to teenage accounts. The two parents speaking to the Senate alongside Raine are suing Character.AI, alleging that the firm’s role-playing AI bots directly contributed to their children’s actions. (A spokesperson for Character.AI told us that the company sends its “deepest sympathies” to the families and pointed us to safety features the firm has implemented over the past year.)

[Read: ChatGPT gave instructions for murder, self-mutilation, and devil worship]

AI firms have acknowledged these problems. In advance of Tuesday’s hearing, OpenAI published two blog posts about teen safety on ChatGPT, one of which was written by the company’s CEO, Sam Altman. He wrote that the company is developing an “age-prediction system” that would estimate a user’s age—presumably to detect if someone is under 18 years old—based on ChatGPT usage patterns. (Currently, anyone can access and use ChatGPT without verifying their age.) Altman also referenced some of the particular challenges raised by generative AI: “The model by default should not provide instructions about how to commit suicide,” he wrote, “but if an adult user is asking for help writing a fictional story that depicts a suicide, the model should help with that request.” But it should not discuss suicide, he said, even in creative-writing settings, with users determined to be under 18. In addition to the age gate, the company said it will implement parental controls by the end of the month to allow parents to intervene directly, such as by setting “blackout hours when a teen cannot use ChatGPT.”

The announcement, sparse on specific details, captured the trepidation and lingering ambivalences that AI companies have about policing young users, even as OpenAI begins to implement these basic features nearly three years after the launch of ChatGPT. A spokesperson for OpenAI, which has a corporate partnership with The Atlantic, declined to respond to a detailed list of questions about the firm’s future teen safeguards, including when the age-prediction system will be implemented. “People sometimes turn to ChatGPT in sensitive moments, so we’re working to make sure it responds with care,” the spokesperson told us. Other leading AI firms have also been slow to devise teen-specific protections, even though they have catered to young users. Google Gemini, for instance, has a version of its chatbot for children under 13, and another version for teenagers (the latter had a graphic conversation with our colleague Lila Shroff when she posed as a 13-year-old).

[From the August 2025 issue: Sexting with Gemini]

This is a familiar story in many respects. Anyone who has paid attention to the issues presented by social media could have foreseen that chatbots, too, would present a problem for teens. Social-media sites have long neglected to restrict eating-disorder content, for instance, and Instagram permitted graphic depictions of self-mutilation until 2019. Yet like the social-media giants before them, generative-AI companies have decided to “move as fast as possible, break as much as possible, and then deal with the consequences,” danah boyd, a communication professor at Cornell who has often written on teenagers and the internet (and who styles her name in lowercase), told us.

In fact, the problems are now so clearly established that platforms are finally beginning to make voluntary changes to address them. For example, last year, Instagram introduced a number of default safeguards for minors, such as enrolling their accounts into the most restrictive content filter by default. Yet tech companies now also have to contend with a wave of legislation in the United Kingdom, parts of the United States, and elsewhere that compel internet companies to directly verify the ages of their users. Perhaps the desire to avoid regulation is another reason OpenAI is proactively adopting an age-estimating feature, though Altman’s post also says that the company may ask for ID “in some cases or countries.”

Many major social-media companies are also experimenting with AI systems that estimate a user’s age based on how they act online. When such a system was explained during a TikTok hearing in 2023, Representative Buddy Carter of Georgia interrupted: “That’s creepy!” And that response makes sense—to determine the age of every user, “you have to collect a lot more data,” boyd said. For social-media companies, that means monitoring what users like, what they click on, how they’re speaking, whom they’re talking to; for generative-AI firms, it means drawing conclusions from the otherwise-private conversations an individual is having with a chatbot that presents itself as a trustworthy companion. Some critics also argue that age-estimation systems infringe on free-speech rights because they limit access to speech based on one’s ability to produce government identification or a credit card.

OpenAI’s blog post notes that “we prioritize teen safety ahead of privacy and freedom,” though it is not clear about how much information OpenAI will collect, nor whether it will need to keep some kind of persistent record of user behavior to make the system workable. The company has also not been altogether transparent about the material that teens will be protected from. The only two use cases of ChatGPT that the company specifically mentions as being inappropriate for teenagers are sexual content and discussion of self-mutilation or suicide. The OpenAI spokesperson did not provide any more examples. Numerous adults have developed paranoid delusions after extended use of ChatGPT. The technology can make up completely imaginary information and events. Are these not also potentially dangerous types of content?

And what about the more existential concern parents might have about their kids talking to a chatbot constantly, as if it is a person, even if everything the bot says is technically aboveboard? The OpenAI blog posts touch glancingly on this topic, gesturing toward the worry that parents may have about their kids using ChatGPT too much and developing too intense of a relationship with it.

Such relationships are, of course, among generative AI’s essential selling points: a seemingly intelligent entity that morphs in response to every query and user. Humans and their problems are messy and fickle; ChatGPT’s responses will be individual and its failings unpredictable in kind. Then again, social-media empires have been accused for years of pushing children toward self-harm, disordered eating, exploitative sexual encounters, and suicide. In June, on the first episode of OpenAI’s podcast, Altman said, “One of the big mistakes of the social-media era was the feed algorithms had a bunch of unintended negative consequences on society as a whole and maybe even individual users.” For many years, he has been fond of saying that AI will be made safe through “contact with reality”; by now, OpenAI and its competitors should see that some collisions may be catastrophic.

What Women’s Baseball Will Look Like

2025-08-27T17:40:29-04:00

The mosquitoes and the National Guard were out, but it was otherwise a perfect day in the capital. Clear and sunny, not too hot: baseball weather. The first pitch was at about 9:30 in the morning. A player waiting in the dugout yammered “Whaddaya say, whaddaya say” before nearly every pitch. Another, after working a long at-bat and winning a walk, celebrated by turning to her teammates and tossing her bat gently toward them with both hands, palms up, like she was presenting them with a gift.

It was a regular workday, a Monday, for the rest of Washington, D.C., but inside Nationals Park, it was the final day of tryouts for the new Women’s Professional Baseball League. This will be the first of its kind since the dissolution of the All-American Girls Professional Baseball League—a wartime entertainment that gave hundreds of women the opportunity to play baseball in front of paying fans, but which fell apart in the early 1950s due to mismanagement and dwindling attendance.

More than 600 players from 10 countries, including Japan, Australia, Canada, and Venezuela—places that have fielded successful teams in the Women’s Baseball World Cup—had reported on the first day of drills and evaluation. The tryouts were led by Alex Hugo, a former player who won a silver medal with the U.S. team during the most recent World Cup and who said in a Monday press conference that the open-tryout format was designed to find “anybody that we would have missed just trying to search ourselves.” Over the weekend, women were evaluated in the batting cages, in fielding drills, and as pitchers, with cuts at the end of each day. The count had been narrowed to just more than 100 for Monday’s doubleheader of scrimmages, which was open to the public. Those who made the final cut in the tryouts will be eligible for a draft in October.

The ceremonial first pitch was thrown out by Maybelle Blair, the 98-year-old elder stateswoman of women’s baseball, who played for the AAGPBL’s Peoria Redwings and now uses a cane made out of a baseball bat. Instantly identifiable by her white bouffant and chunky sunglasses, Blair has been a celebrity for many years, and is often associated with the 1992 movie A League of Their Own, which resurfaced women’s-baseball history in popular memory. “You have no idea the fun I went through when I was playing ball and how I wish that these girls could have the same opportunity,” she said in a press conference afterward. “I never in holy, holy life figured that we would have another league of their own, and here it is.”

A few hundred were people in the stadium, many of them families with young children. Preteen girls who’d come with their parents ate stadium nachos for breakfast and cheered for players who are household names, at least in certain households—Mo’ne Davis, who, 11 years ago, was the first girl to pitch a shutout in the Little League World Series; Alli Schroder, a Canadian pitching phenom who is also a firefighter (a baseball commentator’s dream). One roaming pack of three girls and two boys ran around the stands looking for Kelsie Whitmore, the face of the new league and arguably the most famous woman baseball player in the United States. She was one of the first women to play professional baseball, in a men’s independent league in 2016, and is currently pitching for the Savannah Bananas, the Harlem Globetrotters of baseball. The (mostly male) Bananas play regular baseball, except they also dance and do tricks and comedy bits during the game (and it counts as an out when a fan catches a foul ball).

When Whitmore came up to bat, a mom and daughter seated near me cheered enthusiastically. “Do you know her?” I asked, because many in the stands were there to support family members. “Yeah, who doesn’t know her?” the mom, a New Yorker named Jennifer Montero, replied. “It’s Kelsie Whitmore.” She and her daughter, Edally, had responded to the open call for players, but Edally was only 16 and had been told to come back when she was older. They stayed for the rest of the week anyway to watch. “It’s definitely surreal,” said Edally, who works on her curveball on the weekends in Central Park and plays on her high school’s otherwise all-boys baseball team. “It gives me hope, knowing I’m not working towards nothing.”

The league will start small, with four to six teams. They will play in small ballparks predominantly in the Northeast—places with about 3,000 seats, one of the league’s co-founders, Justine Siegal, told me. These are roughly half the size of those used by lower-level Minor League teams affiliated with Major League Baseball. Still, however modest its beginning, this league is historic: Though I wrote a feature on the history of women’s baseball in the U.S. earlier this year, I was still a little surprised when Whitmore and Davis used the word integration in the press conference, pointing out that the AAGPBL had been whites-only. They’re right. The WPBL, when it starts play in the spring of 2026, will be the first-ever integrated baseball league for women in the U.S.

[From the April 2025 issue: Why aren’t women allowed to play baseball?]

When I spoke with Whitmore after the conference, she rattled off a list of things she hopes to see in the next five years. That would be a full six-month season, a full spring training, maybe a winter league to help accelerate player development. There should be high-school and college baseball for girls in order to create a pipeline of talent, and the women should have salaries that allow them to make baseball their full-time job (a common issue with women’s sports). While playing for the Savannah Bananas, she is also getting a glimpse of the further-off future. “I feel like I’m living two different dreams right now,” she said. “I’m in an environment of playing women’s professional baseball, and then, on top of it, I’m playing in front of sold-out crowds in Major League parks. So, I mean the ultimate goal is we have sold-out crowds for women’s professional baseball.”

In the meantime, she was thrilled by the few days she’d gotten to spend with women who might be her teammates next year. She told me that she feels more like herself and plays more freely “with the girls.” “They’re just a breath of fresh air,” she said. Usually, when this happens—at an international tournament or after an exhibition game—the women have no idea when their next opportunity to play together will be. With a new league on the horizon, that’s over.

What that new league will look like in practice, and how would-be fans will engage with it, is still somewhat of a mystery, but the Savannah Bananas are an interesting parallel. Their goofy theatrics are not to my personal taste, but it’s obvious people like them in part because they feel approachable in a way that Major Leaguers really can’t. During the morning game at the tryouts, players who were scheduled for the second game lounged in the stands among everyone else. At one point, I watched a girl in an Aaron Judge jersey walk up and get an autograph from a WPBL player who was just finishing a hot dog.

Mo’ne Davis signs a baseball during tryouts. (Win McNamee / Getty)

The casualness reminded me of a conversation I had with Kevin Baker, the author of The New York Game: Baseball and the Rise of a New City, earlier this year. We were talking about how a new women’s league might be able to differentiate itself by recapturing some of the old neighborhood spirit of baseball. The Dodgers were just guys who lived in Brooklyn; Mickey Mantle walked to work through Central Park. “Players are so much more aloof now and kind of have to be aloof; I don’t blame them for it,” he said. “But you know, when they could live among us, that was in a way more thrilling.”

That’s one of many ways in which the women’s game might be different. In the stands, I spoke with a group of four players from Vancouver who’d come to the tryouts together and offered various other practical considerations. The women’s league will use metal bats instead of wooden ones. “Realistically, we don’t hit the ball as hard or as fast as men,” Claire Eccles, a pitcher and an outfielder, told me. Metal bats will mean more hits and a faster game, which is what people generally want to see. (Though it’s a new challenge for some of the pitchers who are used to playing with men and throwing to wooden bats.)

Juliette Kladko, a pitcher and first baseman sitting next to Eccles, guessed that the average fastball at the tryouts was probably in the range of 70 to 75 miles an hour. Professional men usually throw in the mid-90s or harder, so women who have played with men their whole life have often focused more on the timing and location of their pitches, the shape of their breaking balls, and what old-timers call the “lost art” of pitching. All four of the Vancouver women had a curveball in their repertoire, and one of them, Eccles, had a knuckleball. The classic curveball is an endangered pitch in velocity-obsessed Major League Baseball, and there are currently no knuckleballers.

The WPBL could offer a looser, more familiar, backyard kind of play, even if it intends to roster elite talent. Not only may the pitching be more painterly; the pitchers will also be the batters, base runners, and defenders. Shohei Ohtani, the Dodgers’ $700 million superstar, is an anomaly and a thrilling novelty because he has continued to pitch and hit at the highest level, even after the practice went completely out of style in the age of the designated hitter. In the WPBL, that would be the norm. Most of the women have been compelled by circumstances and limited opportunities to be super–utility players, and the WPBL teams will probably not even have full-size rosters, so it will remain necessary for women to do it all.

The scrimmages I watched were a bit sloppy at times—lots of hit-by-pitches, lots of defensive errors—but they had exhilarating moments too. On a sharp, bang-bang double play, someone behind me let out a “Hoo, hoo, that was sweet.” After I watched a great play in the outfield, I chatted with two older men in the stands. One of them, Jeff Stewart, told me he’d also gone to watch the Colorado Silver Bullets, a women’s barnstorming team that played for a short time in the 1990s. He was impressed by the WPBL games, he said, and excited for the new league. Obviously, there was room for improvement, but there was a lot of potential. “You saw it!” he said. “That girl in center field just made an outstanding catch.”

The day was generally jubilant, but there was a hum of anxiety in the air. Siegal more than once made a point of saying that the league was going to be built to last and would be around, as she put it, forever. “My grandchild is going to play in this league,” she said in the press conference. Although everyone present certainly wanted that to be true, it doesn’t feel like a given. The first season of the new league will be only four weeks long, followed by a week of All-Star events and two weeks of playoffs, barely a blip on the calendar in comparison with Major League Baseball. During the four weeks of the regular season, each team will play two games a week.

Nobody expected the league to start with 162 games a year, but this seems awfully short—like the season would have hardly begun before it was over already. Montero, the mom who came with her 16-year-old, was dismayed. “Definitely it should be longer, way longer,” she said. “We’ve waited how many years?”

The Culture War Over Nothing

2025-08-15T16:30:00-04:00

“You know the LIBS are seething over this,” Joe Kinsey, an editor at the sports website OutKick, wrote on X while reposting a video of sorority girls doing a choreographed dance. Many of the girls were wearing red-white-and-blue outfits, though some were dressed as hot dogs. They waved American flags in front of a banner that read We Want You Kappa Delta. “Credit to these ladies for pumping out patriotism to kick off the 2025 school year,” Kinsey wrote.

It wasn’t only the display of patriotism that supposedly made liberals seethe. “The purple hair lesbians have to be furious that SEC sororities ARE BACK,” Kinsey wrote while reposting another sorority-dance video. This one had no clear Americana element aside from the matching trucker hats all of the dancing girls were wearing. Kinsey’s two posts were viewed nearly 40 million times.

Many other such videos have been shared on X in the past couple of weeks, as sororities have begun recruiting for the new school year. The videos come from TikTok, where sorority dance videos have long been popular. But they’ve been presented on X with a new gloss: Democrats, liberals, and leftists are enraged by pretty, mostly white young women who are dancing happily. It drives them up the wall when a woman is blond! Do not let a liberal see a woman smiling while wearing a short denim skirt.

The only thing that is missing is evidence of seething libs. Search around social media, and you might be surprised how difficult such reactions are to find. In fact, I couldn’t find a single one. When I asked Kinsey where he got the idea that people were angry about the sorority-recruitment videos, he didn’t point me to any specific examples. He noted that many people replied to his posts saying that they weren’t mad about the TikTok dances. But, he said, “I don’t believe that.”

By now, this is all familiar. Recall the recent controversy over an American Eagle ad starring Sydney Sweeney, in which the actress hawked denim jeans by making a pun about her genes. A small number of people on social media did get very angry, and posted about how the ad sounded like a eugenics dog whistle. Their reaction was then amplified by right-wing commentators eager to make the point that the left hates hot women. The fact that the situation involved Sydney Sweeney, a celebrity who had already been evoked in culture-war debates in the past, drove even more attention. It turned into a full-blown news cycle. (I am confident my grandmother heard about this.)

[Read: The discourse is broken]

In both cases, this burst of bizarre posting is less a story about American politics than it is a story about social media and, specifically, X. Whatever else you may say about Elon Musk’s platform, it is the best place to watch a fake drama unfold.

Both of the videos that Joe Kinsey shared—of the girls with the flags and the girls with the trucker hats—were originally posted on their respective sororities’ TikTok accounts. But the versions he shared had been uploaded to X by what appears to be an account called “Calico Cut Pants,” which seemingly exists to move short-form videos from one platform to another. The account follows no one and is named after a sketch from the Tim Robinson Netflix show I Think You Should Leave. Other sorority dance videos have been pulled from TikTok and posted by an account called “Big Chungus,” which also posts almost nothing but videos from other sites, paired with incendiary rhetoric.

Accounts like these can bring in money by driving engagement on X, thanks to a revenue-sharing program that debuted after Musk took over the site. Both Big Chungus and Calico Cut Pants have Premium badges, which means they can get paid for generating activity, including likes and replies. According to X’s Creator Revenue Sharing guidelines, the company maintains some discretion in calculating the true “impact” of posts. For instance, engagement from other paid accounts is worth more than engagement from an unpaid account. It stands to reason that the best way to make money is to elicit some reaction to your content from the people who enjoy X enough to pay for it. Social media is replete with political outrage, and playing to either a liberal or conservative audience is likely to draw attention. (Certainly, plenty of accounts decrying MAGA values, real and exaggerated, exist.) But X, in particular, is a much more right-coded platform than it was a few years ago, and it makes sense to pander to the home crowd.

Consider “non aesthetic things,” an account that has 4.9 million followers on X, all from posting short-form videos—sometimes relatable, sometimes nostalgic, generally just mind-numbing. Its bio links to an Instagram page that is full of ads for the gambling company Stake. (None of these accounts responded to requests for an interview.) The non aesthetic things account shared a video of sorority girls at Arizona State University who were performing in jean shorts, most of them quite short, and cowboy boots. The X caption makes reference to “their JEANS”—a subtle nod to the Sydney Sweeney panic. This pairing of footage and wink was a solid bet to produce a big reaction.

Given all the attention the Sweeney dustup received, returning to it is logical for engagement farmers. “BREAKING,” wrote a pro-Trump account called “Patriot Oasis” that almost exclusively posts short-form videos, “Sorority at the University of Oklahoma wearing ‘Good Genes’ is going VIRAL showcasing pure American beauty. Liberals are OUTRAGED online.” The caption suggested that the sorority is participating in some kind of activist response to the villainization of Sydney Sweeney, though there is no reason to believe that. The girls in the video never say anything about politics, Sydney Sweeney, genes, or even jeans. The sorority has been making similar dance videos for years.

Nevertheless, the right-wing influencer Charlie Kirk reposted Patriot Oasis to his 5.1 million followers and asked, “Do you see the difference between conservative and liberal women?” Underneath his post, a Community Note generated by other users pointed out that the video doesn’t reveal whether the women are conservative or not. But that hardly mattered. Many others made the same argument in the replies to Kirk’s post, driving up engagement. Although the original post has since been deleted, Kirk’s repost has more than 3.8 million views.

[Read: Everyone wants to be a hot, anxious girl on Twitter]

Sorority dances worked well on social media even before they were inserted into a fake culture-war debate, because they are briefly hypnotic due to the weirdness of so many people moving in the same way while wearing such similar outfits. They offer the muted thrill of a flash mob. But plucked from their original context, they offer more. Someone finds them and puts them on X with just a phrase or two of framing and they blow up.

People watch the videos of young women dancing and gleefully share them, writing, for example, “nothing is more triggering to leftists,” and “at what point do you just give up if you’re a lib?” and “America is BACK and Democrats hate it.” There is no need to point to an actual instance of a leftist or lib or Democrat being triggered. It is easy enough to imagine how triggered they are.

The Epstein ‘Client List’ Will Never Go Away

2025-08-07T11:07:59-04:00

Jeffrey Epstein’s “client list” is the conspiracy theory that may never die. A secret document detailing all of the elite clients that Epstein allegedly sex-trafficked minors to—it’s something of a grail for QAnon adherents, TMZ watchers, and serious news readers alike. There is no proof that such a thing exists.

Yet President Donald Trump himself suggested that it did during his campaign, and pledged to release it before a disastrous backtrack from the Department of Justice last month. Now, in a poll released Monday, nearly two-thirds of Americans said they believe that the Trump administration is hiding something, and 71 percent said they still believe that the list is real. Meanwhile, Representative Marjorie Taylor Greene has demanded that the list be released, Democrats are pushing the narrative that the Trump administration is orchestrating a cover-up, and yesterday the House subpoenaed the DOJ for additional files related to the case.

To be clear, many unanswered and valid questions remain about Epstein. Before his death, he was charged with trafficking and abusing, as it read in the indictment, “a vast network” of dozens of underage girls. Many still wonder why he was permitted to carry on with his crimes for so long, whether other people who were complicit in them have escaped justice, and how much President Trump may have known while the two were friends. Trump’s name reportedly appears in files that have been redacted by the FBI, though he has repeatedly denied personal knowledge of Epstein’s crimes and says their relationship ended in 2004.

[David A. Graham: Donald Trump doesn’t want you to read this article]

The specific idea of a client list, though, has taken on a life of its own. No one can demonstrate that the list doesn’t exist, so people will continue to insist that it does—that it is being kept from them. There’s a certain logic to their belief, because a similar document has been seen already. In 2015, Gawker published Epstein’s address book, which was full of names of celebrities and politicians. He apparently kept meticulous records and liked putting all of his famous contacts together in one place. And so the idea of a client list feels plausible to many people because they’ve had a mental image of it for 10 years now.

Moreover, Trump has created a “where there’s smoke there’s fire” effect in the past several weeks. The president has vacillated among suggesting that he has no obligation to talk about Epstein, speculating that political foes may have fabricated parts of the Epstein file, attempting to placate his supporters by ordering the release of grand-jury testimony about the case (which cannot be unsealed, a federal judge ruled), and deflecting (“you ought to be talking about Bill Clinton”).

There’s a useful parallel between the government’s handling of the Epstein case and its investigation into the John F. Kennedy assassination. That assassination, of course, launched a million conspiracy theories: Most Americans still believe that the shooter, Lee Harvey Oswald, did not act alone. One theory holds that the CIA was somehow involved, which has led people to search for hidden evidence within the government’s own records—much as we’ve seen with the Epstein case.

In 1967, Jim Garrison, the district attorney of New Orleans, ended up going down this road. He was re-investigating the case after receiving tips that Oswald, a New Orleans native, had worked with locals in a plot to kill the president. Long and complicated story short, Garrison would eventually subpoena CIA Director Richard Helms, demanding that he produce a photograph that purportedly showed Oswald with a CIA officer in Mexico City in 1963—cementing a link between the killer and the intelligence agency.

There was only a slim reason to think such a photo might exist. Garrison was extrapolating from an existing controversy over a photo that the CIA had provided to the Warren Commission years before. That photo showed an unknown man in Mexico City; it was labeled as a photo of Oswald but was clearly not him. Garrison’s theory was that there had been a swap. “It’s perfectly clear that the actual picture of Oswald and his companion was suppressed and a fake photo substituted,” he said. The government had no way to prove that he was wrong—to prove that there was no such photo. Garrison took his accusations all the way to a highly publicized trial in 1969. His theory of the case fell apart in court for unrelated reasons, but his many notions linger to this day. (He is the hero of the 1991 blockbuster film JFK.)

The Kennedy assassination still features many unknowns, and information is still being released about it in drips and drabs—previewing, perhaps, the future of disclosure around the Epstein case. Last month, the CIA released assassination files that researchers had been requesting for more than 20 years. They pertained to a specific CIA officer who some think may have known or worked with Oswald in New Orleans. In the 1970s, the same CIA officer was assigned to work with the House Select Committee on Assassinations and help them in their re-investigation of Kennedy’s death. He was using a different name by then, and the committee did not know it was the same person. He blatantly deceived Congress and actually thwarted their efforts to understand whatever had happened in New Orleans. The latest batch of files still didn’t reveal a direct connection between this officer and Oswald, but that hasn’t put the issue to bed.

[Read: Conspiracy theorists are turning on the president]

That the CIA maintained its secrecy around the officer for decades is what has made curiosity linger. The historian Gerald Posner was one of the public figures (along with the novelist Don DeLillo and the writer Norman Mailer) who’d signed an open letter asking for the release of these files back in 2003, a decade after he wrote a definitive book affirming the theory that Oswald acted alone. He recently told me that he’s disgusted with the CIA for taking so long to provide them—not because he thinks they shed new light on the Kennedy assassination but for just the opposite reason. He thinks they really don’t, but that hiding them encourages people to speculate ever more darkly. The CIA drags its feet, and when the documents are finally released, they usually have “nothing to do with the assassination,” Posner said. “But it’s often too late to explain that.”

This dynamic—in which defensiveness and reflexive secrecy lead to prolonged struggles over information that may or may not be important—has been a recurring problem throughout modern U.S. history. In her 2008 book, Real Enemies: Conspiracy Theories and American Democracy, World War 1 to 9/11, the historian Kathryn Olmsted argues that selective opacity is one of the key reasons that Americans distrust their government. The passage of the Freedom of Information Act in 1966 democratized access to information, she argues, yet also left citizens baffled and frustrated when documents were refused to them or granted only with heavy redactions. The government’s “ambivalence” about providing information “sometimes had the effect of frightening citizens rather than reassuring them,” Olmsted writes.

There are good reasons that not all of the Epstein files can be released—chief among them, the privacy of victims—but Americans are not wrong to think the government is being less transparent than it could be. The administration could release more than it has, which Congress is currently pressuring it to do. Within that context, why would people believe Trump or the FBI when they say that a client list doesn’t exist? I posed this question to Mark Fenster, a professor at the University of Florida’s law school who often writes about government transparency and conspiracy theories. Can you ever convince people that there is no list? “No, you can’t,” he said. “You can’t convince people that all of the pertinent JFK-assassination documents have been released. You can’t convince people who believe otherwise that all the truth is out on Jeffrey Epstein.” (Especially because it currently isn’t.) “That’s just a flat no,” he went on. “Rarely do I say flat nos, but that’s just a flat no.”

Like the Epstein case, Kennedy-assassination skepticism demonstrates two opposing impulses. The first, to speculate wildly. The second, to doggedly pursue more and better information, sometimes so stubbornly that it approaches irrationality in itself. These past few weeks have also brought to mind the Kennedy researcher Harold Weisberg, whose early books were a countercultural phenomenon and who was known for his diligent, insistent filling of FOIA requests. He wanted a specific report that he thought must exist about the spectrographic testing used on the Dallas crime-scene bullets; he was told that the FBI had looked for such a report and couldn’t find anything. He appealed four times before the D.C. Circuit ruled in 1983 that he had to stop. The decision stated that if an agency could prove it had conducted a thorough search for the requested material, it did not also have to prove the negative—that the material never existed or had previously been destroyed. Yet, of course, the court couldn’t compel him to stop wondering.

Nobody can make Americans stop wondering about a “client list” either. It can’t stay on the front page indefinitely, but people won’t forget about it. Epstein will become part of the American cultural backdrop, like Hunter Biden’s computer, 9/11 trutherism, Kennedy, chemtrails, Roswell, and QAnon. At certain times, such conspiratorial thinking and refusal to accept the evidence will become dangerous—people will spin up fantasies that result in acts of defamation or threats of violence. At other times, it will just be part of the daily chatter.

Conspiracy Theorists Are Turning on the President

2025-07-10T17:41:11-04:00

This article was featured in the One Story to Read Today newsletter. Sign up for it here.

The Trump administration had promised a bombshell. Americans, many of whom had spent years wondering over the unknowns in the Jeffrey Epstein case, would finally get their hands on the secret files that would explain it all. What really happened when the accused sex trafficker died in jail back in 2019? And who was on his “client list”—a rumored collection of famous and powerful people who participated in Epstein’s crimes?

In a September 2024 interview on the Lex Fridman Podcast, Donald Trump suggested that he would release the list if reelected. “Yeah, I’d be inclined to do the Epstein; I’d have no problem with it,” Trump said. He indulged speculation about Epstein after his reelection as well. In February, the White House hosted a collection of MAGA-world influencers and gave them binders full of heavily redacted Epstein-related documents labeled Phase 1, suggesting more to come.

The Trump administration has been unusually focused on messaging about such information, making a show of pulling the curtain back on supposed secrets. Trump similarly promoted the release of further documents related to the John F. Kennedy assassination, along with records on the killings of Martin Luther King Jr. and Senator Robert F. Kennedy. In an executive order signed this January, the administration framed these efforts as “PROVIDING AMERICANS THE TRUTH.” At an April hearing on those files, Nancy Mace, a Trump ally and representative from South Carolina, brought up the so-called Epstein list. In a meandering statement, she spoke about her desire to see documents regarding Epstein, as well as Hunter Biden’s laptop and the origins of the coronavirus. All have been recurring internet fascinations among Trump’s supporters. “Sunshine literally is the best medicine,” Mace argued.

A personal wish list of coveted secrets is not exactly the same thing as a principled call for government transparency. But the two are easy to conflate and can have some incidental overlap, which can be politically useful. The promise of previously withheld revelations has allowed Trump to frame himself as an outsider fighting on behalf of voters who have been kept in the dark by the establishment. The catch is that once he was back in office, he was put in the awkward position of having to deliver.

On Monday, the FBI released a memo saying that it had reviewed all of its files on Epstein and that it does not plan to release more after all; there will be no Phase 2. According to the FBI, only a “fraction” of the remaining material would have become public if Epstein had lived to go to trial, because it includes “a large volume” of illegal content involving underaged victims of sexual abuse—in other words, material that cannot be released to the public. The memo also noted, in one breezy paragraph, that the bureau’s review had uncovered neither a client list nor evidence “that Epstein blackmailed prominent individuals as part of his actions.” There will be no new investigation against “uncharged third parties,” the memo said. This has come as a shock to a group of people who have long bought into the idea that Trump would one day unmask an evil ring of Democrats and liberal-coded celebrities.

Anna Paulina Luna, a representative from Florida and the chair of the Task Force on the Declassification of Federal Secrets, which facilitated the recent document releases regarding JFK, told me that she will be asking the Department of Justice to authorize the release of more Epstein details anyway. “I think the American people still have questions and there is stuff that they can release,” she said. She didn’t comment specifically on the existence of a client list and said she didn’t yet know exactly what kind of documents the FBI might still have (clarifying that she agreed that the bureau should not release any private details about victims or child-sexual-abuse material).

In the meantime, the about-face on the Epstein files is splintering MAGA world, and many Trump allies are feeling betrayed and unmoored. “No one believes there is not a client list,” wrote Marjorie Taylor Greene, the representative from Georgia who has avidly promoted QAnon conspiracy theories. “This is a shameful coverup to protect the most heinous elites,” one of the influencers who went to the White House in February, Rogan O’Handley (who goes by “DC Draino”), told his more than 2 million X followers on Monday. Longtime Trump loyalists, including the conspiracy theorist Alex Jones, started sharing a meme on Monday that depicted a bunch of cartoon lizard people laughing about having pulled one over on the unsuspecting public yet again. Strange—some readers may be old enough to remember when it was Hillary Clinton and other Democrats who were the shadowy reptilian elite, secretly shedding their human skin whenever out of public sight.

Significant ire has been directed at Attorney General Pam Bondi, who responded to a question about a client list in February by saying it was “sitting on my desk right now to review.” During a press conference on Monday afternoon, White House Press Secretary Karoline Leavitt said that Bondi had actually been referring to “the entirety of all the paperwork” on Epstein and not to a specific document. Shortly thereafter, the online crowd began questioning why Leavitt had not been wearing her usual cross necklace at the briefing—a sign, perhaps, that she was lying and didn’t want to do so in front of God (to paraphrase the posts, which were mostly ruder than that). When I asked Luna if Trump’s supporters had a right to feel frustrated, she deflected the question, saying, “I can’t speak for people on the internet or the president. What I can say is President Trump is on the cusp of negotiating a permanent cease-fire with Israel and Hamas in Gaza. This is overshadowing the amount of success the administration has had in that sense.”

Yet this is undeniably a turning point for the highly online among Trump’s base. The story of the client list had effectively morphed into a more palatable and plausible version of the QAnon conspiracy theory. As does QAnon, it features a secret ring of evildoers, though it doesn’t have certain ostentatious elements of that conspiracy (no harvesting blood). But both theories encourage people to disbelieve everything the government tells them. Until now, Trump and his appointees were positioned as exceptions to that rule—the deal was that if they got back into power, they would reveal all.

[From the June 2020 issue: The prophecies of Q]

Mark Fenster, a professor at the University of Florida’s law school who has written about government transparency and conspiracy theories, observed to me that, with his administrative appointments, Trump had made implicit promises to his supporters. “He specifically nominated people for high-level positions who have been engaged in conspiracy theories for the past five-plus years,” Fenster pointed out. For instance, FBI Director Kash Patel and Deputy Director Dan Bongino have offered wild theories about the Epstein case in the past—Patel once suggested that the FBI may be covering up evidence to protect unnamed elites, while Bongino said he’d heard a rumor that Epstein was a foreign intelligence agent.

Now the conspiracy is mutating again to fit the administration’s reversal. “To hear Pam Bondi and to hear Kash Patel and Don Bongino saying there is no list—you’re going to say, ‘Well, they must be part of the conspiracy too,’” Fenster suggested, which is certainly one avenue people have gone down. Because the FBI’s memo coincided roughly with a diplomatic visit of Israeli Prime Minister Benjamin Netanyahu to the White House, others have started suggesting that Epstein was secretly a Mossad agent (a claim often expressed with anti-Semitic rhetoric). Alex Jones, who was initially furious about the FBI memo, has since speculated that Trump has actually taken “control” of the alleged list and is using it to blackmail the “deep state” behind the scenes.

Of course, some have started picking apart the FBI memo itself. It concluded with links to two videos of a hallway in the Metropolitan Correctional Center where Epstein had been held, showing that nobody went into his cell the night of his death. Viewers quickly noticed that the clock in the corner of the video skips from 11:59:00 to 12:00:00, which suggested to them that a minute of footage was missing.

On Tuesday afternoon, when a reporter attempted to ask Bondi about the foreign-intelligence theory and the video-clock issue, Trump cut in. “Are you still talking about Jeffrey Epstein?” he asked incredulously. “This guy’s been talked about for years. You’re asking—we have Texas, we have this, we have all of the things, and are people still talking about this guy, this creep? That is unbelievable.” Bondi said she didn’t mind answering the question, but Trump went on. “I can’t believe you’re asking a question on Epstein at a time like this where we’re having some of the greatest success and also tragedy with what happened in Texas,” he said, referring to the flooding that has killed at least 120 people.

Eventually, he waved for Bondi to go ahead. She told the reporter she had no knowledge of Epstein being an agent, then explained that the video hadn’t been doctored and that the clocks on the outdated cameras in the Metropolitan Correctional Center always jump ahead as they approach midnight. From what I saw, hardly anyone online was buying this explanation, which comes as no surprise. Trump and his administration invited conspiracy theories into the White House. Now they’re going to have a hard time getting them out.

Why Would the Trump Family Want to Run a Phone Company?

2025-06-18T14:33:53-04:00

The Trumps are doing phones now. This week, the Trump Organization announced its own cellphone service called Trump Mobile, as well as a gold-colored smartphone called the T1, which will purportedly be manufactured in the United States and retail for $499. It is available for preorder now and will supposedly ship in August or September, though one reporter who attempted to buy the device was left feeling unsure: His card was charged $64.70 instead of the full $100 down payment, and he was never asked to provide a shipping address.

What other details do you need? “Trump Mobile is going to revolutionize kind of, you know, cellphones,” Eric Trump, the president’s son and an executive for the Trump Organization, said on Fox Business. According to Trump Mobile’s Terms of Use page, its service will be “powered by” Liberty Mobile, which itself runs on T-Mobile and uses the clever tagline “Let Freedom Ring.” Other marketing materials confuse the issue by suggesting that Trump Mobile works with all three major carriers. The phone plan will cost $47.45 a month, which is somewhat expensive for this type of service but makes sense numerologically with Trump’s brand (47th and 45th president).

To be clear, Trump is not building out his own networking infrastructure. Trump Mobile will be a mobile virtual network operator (or MVNO). These essentially buy service from major providers such as T-Mobile and AT&T at a discounted, wholesale rate, and then sell that service to customers who are comfortable with making various compromises in exchange for a much lower bill than they’d have with the mainstream carriers. This is about the extent of the available details. The Trump Organization did not return my request for additional information about where the phone would be made and by whom, nor did it answer my question about whether the phone currently exists physically. (The images on the website appear to be not photographs, but questionable mock-ups—the camera is depicted without a flash, as noted by The Verge.) I also asked the Trump Organization whether the Trump family faces a conflict of interest in entering an industry that is regulated by the Federal Communications Commission, an agency led by presidential appointees; no response.

But I was most interested in my unanswered question about why the Trump Organization would want to be involved in the telecom industry at all. To some extent, the answer is obvious: The Trumps are involved in such a sprawling array of moneymaking endeavors, it would make more sense to ask whether there are any they would not consider trying. They’ve done quite a bit in the tech sector already, between NFTs, memecoins, a social-media platform, and other fascinating ventures.

[Read: The Trump sons really love crypto]

Still, the choice is curious, if only because operating a cellphone service seems so boring and unglamorous. It’s also funny timing: Last week, the actors Jason Bateman, Will Arnett, and Sean Hayes, who co-host the super-popular podcast SmartLess, announced SmartLess Mobile, a discount phone plan that also relies on T-Mobile. That move was not well explained by its participants, either. In an interview with People about the move, Bateman said twice that most people listen to podcasts on phones, and therefore the telecom industry is a logical one for podcasters to enter. “It just kind of organically shaped into something that really made sense for us to try,” he added.

Did it?

The celeb phone companies remind me, a little, of the ISP that David Bowie launched in 1998, which for $19.95 a month offered “full uncensored” internet access, Bowie-themed chat rooms, and a coveted “@davidbowie.com” email address. That service lasted for eight years, which is pretty impressive, but it was more of a highly laborious artistic experiment and act of fan service than an effort to maintain and profit from digital infrastructure long term.

Today’s businesspeople appear to be more directly inspired by the actor Ryan Reynolds’s fortuitous investment in Mint Mobile, another MVNO, which sold for more than $1 billion in 2023. What they’re doing is a step further than what he did, because they’re not just investing in an existing phone company: The Trump Organization and the SmartLess guys are putting their names on something new. The question, then, is: Why would phone companies suddenly appeal to the type of people who might otherwise put their names on bottles of tequila or pickleball paddles or what have you?

I emailed Steffen Oefner, a vice president at Magenta Telekom, the Austrian iteration of T-Mobile (MVNOs are more common in Europe), to ask him. “Interessent point,” he replied. “One answer is … because they can.” The MVNO industry now has a number of middleman companies that will do the work of negotiating with a network and then allow brands or influencers to simply put their name on a ready-made product, he explained. Setting up an MVNO is significantly cheaper than it was 10 years ago. “We do expect more celebrity brands or fan-base MNVOs to appear in the mobile market,” he said. To add to my list, he gave the example of LariCel, a phone company in Brazil affiliated with the actor Larissa Manoela (who has more than 53 million followers on Instagram), which refers to its customers as LariLovers.

After reviewing the list of personalities who appeared at a recent MVNO conference held in Vienna, I found James Gray of Graystone Strategy, which consults with clients in the MVNO space. He agreed with Oefner about the ease of starting an MVNO and also pointed to the invention of digital SIM cards, or eSIMS, which enable people to switch to a new phone plan instantly, without having to wait for a little piece of plastic to be shipped to them. “Now we’re in a digital world,” he said.

[Read: The Trump posts you probably aren’t seeing]

This general point had multiple implications. Previously, he said, companies such as T-Mobile would have preferred to partner with retail companies or banks, enticing new customers by offering them special deals on products or services they were already using. Today, a digital brand such as that of “an influencer or someone running a podcast” can also sell a service, maybe by saying that it represents their values or that it comes with access to a community. “Trump would be a relatively famous brand,” he noted. As another example, he pointed to FC Barcelona, which recently started offering an MVNO called Barça Mobile to its many, many super-enthusiastic fans as a way to be even more intensely involved with the club (while also receiving cheap phone service).

The SmartLess guys are pitching their new venture by saying that a lot of people currently pay for more cellphone data than they actually use, given that they are actually connected to Wi-Fi most of the time (suggesting, I suppose, a customer base that is often either at home or in an office). The Trump plan will offer roadside assistance and access to a telehealth service (suggesting, I suppose, a customer base that is older or generally accident-prone). In the U.S., other politics-themed MVNOs also already exist—the California-based Credo Mobile puts some of its profits into left-wing causes, while the Texas-based Patriot Mobile puts some of its profits into right-wing causes. (The latter identified itself as a trailblazer of “the Red Economy” in a press release congratulating Trump Mobile on its launch.)

Gray concluded that the appeal of the phone business to the celebrities was obvious. “The difference between this and, say, a celebrity vodka is this is recurring revenue,” he told me. “People sign up and they pay a subscription to you every month.” (That was also the case with Rihanna’s underwear membership, though people did eventually get upset about it.) And of course, he’s right—that is the big difference. That is why a famous person would want to run a phone company. We’re in a digital world now. How lucky.

Why Are There So Many ‘Alternative Devices’ All of a Sudden?

2025-05-17T07:00:29-04:00

On a recent commute to work, I texted my distant family about our fantasy baseball league, which was nice because I felt connected to them for a second. Then I switched apps and became enraged by a stupid opinion I saw on X, which I shouldn’t be using anymore due to its advanced toxicity and mind-numbing inanity. Many minutes passed before I was able to stop reading the stupid replies to the stupid original post and relax the muscles of my face.

This is the duality of the phone: It connects me to my loved ones, and sometimes I think it’s ruining my life. I need it and I want it, but sometimes I hate it and I fear it. Many people have to navigate this problem—and it may be at its worst for parents, who’ve recently been drowned in media suggesting that smartphones and social media might be harming their children’s mental health, but who also want their kids to enjoy technology’s benefits and prepare themselves for adult life in a digital age.

[Read: No one knows exactly what social media is doing to teens]

It was with this tension in mind that I rode a train last week to the town of Westport, Connecticut. There, a parent-led group called OK to Delay had organized an “Alternative Device Fair” for families who wanted to learn about different kinds of phones that were intentionally limited in their functionality. (There would be no frowning at X with these devices, because most of them block social media.) Similar bazaars have been popping up here and there over the past year, often in the more affluent suburbs of the tristate area. Westport’s fair, modeled after an event held last fall in Rye, New York, was set up in a spacious meeting room in the most immaculate and well-appointed public library I’ve ever seen. When I arrived, about 30 minutes after the start of the four-hour event, it was bustling. The chatter was already at a healthy, partylike level.

The tables set up around the room each showed off a different device. One booth had a Barbie-branded flip phone; another was offering a retro-styled “landline” phone called the Tin Can. But most of the gadgets looked the same—generic, rectangular smartphones. Each one, however, has its own special, restricted app store, and a slew of parental-control features that are significantly more advanced than what would have been available only a few years ago. One parent showed me her notepad, on which she was taking detailed notes about the minute differences among these phones; she planned to share the information with an online group of parents who hadn’t been able to come. Another mom told me that she’d be asking each booth attendant how easy it would be for kids to hack the phone system and get around the parent controls—something you can see kids discussing openly on the internet all the time.

A couple of years ago, I explored the “dumb phone” trend, a cultural curiosity about returning to the time before smartphones by eschewing complex devices and purchasing something simpler and deliberately limited. One of the better phones I tried then was the Light Phone II, which I disliked only because it was so tiny that I constantly feared that I would break or lose it. At the library, I chatted with Light Phone’s Dan Fox, who was there to show people the latest version of the device. The Light Phone III is larger and thicker and has a camera, but it still uses a black-and-white screen and prohibits web browsing and social-media apps. He told me that it was his third alternative-device event in a week. He’d also been to Ardsley, a village in New York’s Westchester County, and to the Upper East Side, in Manhattan. He speculated that kids like the Light Phone because it doesn’t require all the rigmarole about filters and settings and parents. It was designed for adults, and therefore seems cool, and was designed in Brooklyn, which makes it seem cooler. (Fox then left early to go to a Kendrick Lamar concert with his colleagues.)

[Read: Phones will never be fun again]

The crowded room in Westport was reflective of the broad concern about the effect that social media may have on children and teenagers. But it was also a very specific expression of it. Explaining the impetus for hosting the marketplace, Becca Zipkin, a co-founder of the Westport branch of OK to Delay, told me that it has become the standard for kids in the area to receive an iPhone as an elementary-school graduation present. One of her group’s goals is to push back on this ritual and create a different culture in their community. “This is not a world in which there are no options,” she said.

The options on display in Westport were more interesting than I’d thought they were going to be. They reflected the tricky balancing act parents face: how to let kids enjoy the benefits of being connected (a chess game, a video call with Grandma, a GPS route to soccer practice, the feeling of autonomy that comes from setting a photo of Olivia Rodrigo as your home-screen background) and protect them from the bad stuff (violent videos, messages from creeps, the urge to endlessly scroll, the ability to see where all of your friends are at any given time and therefore be aware every time you’re excluded).

Pinwheel, an Austin-based company, demonstrated one solution with a custom operating system for Android phones such as the Google Pixel that allows parents to receive alerts for “trigger words” received in their kids’ texts, and lets them read every message at any time. As with most of the others demonstrated at the fair, Pinwheel’s custom app store made it impossible for kids to install social media. During the demo, I saw that Pinwheel also blocked a wide range of other apps, including Spotify—the booth attendant told me and a nearby mom that the app contains “unlimited porn,” a pronouncement that surprised both of us. (According to him, kids put links to porn in playlist descriptions; I don’t know if that’s true, but Spotify did have a brief problem with porn appearing in a small number of search results last year.) The app for the arts-and-crafts chain Michaels was also blocked, for a similar but less explicit reason: A red label placed on the Michaels app advised that it may contain a loophole that would allow kids to get onto unnamed other platforms. (Michaels didn’t respond to my request for comment, and Spotify declined comment.)

Beyond the standard suite of surveillance tools, many of the devices are also outfitted with AI-powered tools that would preemptively censor content on kids’ phones: Nudity would be blurred out and trigger an alert sent to a parent, for instance; a kid receiving a text from a friend with a potty mouth would see only a series of asterisks instead of expletives.

“The constant need to be involved in the monitoring of an iPhone is very stressful for parents,” Zipkin told me, referring to the parental controls that Apple offers, which can become the focus of unceasing negotiation and conflict between kids and their guardians. That is part of these alternative devices’ marketing. Pinwheel highlights the helping hand of AI on its website: “Instead of relying on parents to manually monitor every digital interaction (because who has time for that?), AI-driven tech is learning behaviors, recognizing risks, and proactively keeping kids safe.”

The story was similar at other tables. Gabb, a Lehi, Utah, company, offers a feature that automatically shuts down video calls and sends notifications to parents if it detects nudity. The AI still needs some work—it can be triggered by, say, a person in a bathing suit or a poster of a man with his shirt off, if they appear in the background of the call. Gabb also has its own music app, which uses AI and human reviewers to identify and block songs with explicit language or adult themes. “Taylor Swift is on here, but not all of Taylor Swift’s music,” Lori Morency Kun, a spokesperson for the company, told me.

At the next booth, another Utah-based company, Troomi, was demoing a system that allows parents to set content filters for profanity, discussions of violence, and “suggestive” chitchat, on a sliding scale depending on their kid’s age. The demonstrator also showed us how to add custom keywords to the system that would also be blocked, in case a parent feels that the AI tools are not finding everything. (“Block harmful content BEFORE it even has the chance to get to your kiddo!” reads a post on the company’s chipper Instagram account.)

Across the room, Bark, an Atlanta-based company that started with a parental-control app and then launched its own smartphone, offered yet another nice-looking slab with similar features. This one sends alerts to parents for 26 possible problems, including signs of depression and indications of cyberbullying. I posed to the booth attendant, Chief Commercial Officer Christian Brucculeri, that a kid might joke 100 times a day about wanting to kill himself without having any real suicidal thoughts, an issue Brucculeri seemed to understand. But false positives are better than missed negatives, he argued. Bark places calls to law enforcement when it receives an alert about a kid threatening to harm themselves or others, he told me, but those alerts are reviewed by a human first. “We’re not swatting kids,” he said.

Although everybody at the library was enormously polite, there is apparently hot competition in the alternative-device space. Troomi, for instance, markets itself as a “smarter, safer alternative to Pinwheel.” Pinwheel’s website emphasizes that its AI chatbot, PinwheelGPT, is a more useful tool than Troomi’s chatbot, Troodi—which Pinwheel argues is emotionally confusing for children, because the bot is anthropomorphized in the form of a cartoon woman. Bark provides pages comparing each of these competitors, unfavorably, with its own offering.

Afterward, Zipkin told me that parents had given her varied feedback on the different devices. Some of them felt that the granular level of monitoring texts for any sign of emotional distress or experimental cursing was over-the-top and invasive. Others were impressed, as she was, with some of the AI features that seem to take a bit of the load off of parents who are tired of constant vigilance. Despite all the negative things she’d personally heard about artificial intelligence, this seemed to her like a way it could be used for good. “Knowing that your kids won’t receive harassing or bullying material or sexual images or explicit images, or anything like that, is extremely attractive as a parent,” she told me. “Knowing that there’s technology to block that is, I think, amazing.”

Of course, as every parent knows, no system is actually going to block every single dangerous, gross, or hurtful thing that can come in through a phone from the outside world. But that there are now so many alternative-device companies to choose from is evidence of how much people want and are willing to search for something that has so far been unattainable: a phone without any of the bad stuff.

We’re Back to the Actually Internet

2025-05-05T11:24:31-04:00

Updated at 1:53 p.m. ET on May 5, 2025

One Friday in April, Meta’s chief global affairs officer, Joel Kaplan, announced that the process of removing fact-checking from the American versions of Facebook, Threads, and Instagram was nearly complete. By the following Monday, there would be “no new fact checks and no fact checkers” working across these platforms in the U.S.—no professionals marking disinformation about vaccines or stolen elections. Elon Musk, owner of X—a rival platform with an infamously permissive approach to content moderation—replied to Kaplan, writing, “Cool.”

Meta, then just called Facebook, began its fact-checking program in December 2016, after President Donald Trump was first elected and the social network was criticized for allowing the rampant spread of fake news. The company will still take action against many kinds of problematic content—threats of violence, for example. But it has left the job of patrolling many kinds of misinformation to users themselves. Now, if users are so compelled, they can turn to a Community Notes program, which allows regular people to officially contradict one another’s posts with clarifying or corrective supplementary text. A Facebook post stating that the sun has changed color might receive a useful correction, but only if someone decided to write one and submit it for consideration. Almost anyone can sign up for the program (Meta says users must be over 18 and have accounts “in good standing”), making it, in theory, an egalitarian approach to content moderation.

Meta CEO Mark Zuckerberg has called the pivot on misinformation a return to the company’s “roots,” with Facebook and Instagram as sites of “free expression.” He announced the decision to adopt Community Notes back in January, and explicitly framed the move as a response to the 2024 elections, which he described as a “cultural tipping point towards once again prioritizing speech.” Less explicitly, Meta’s shift to Community Notes is a response to years of being criticized from both sides of the aisle over the company’s approach to misinformation. Near the end of his last term, Trump targeted Facebook and other online platforms with an executive order accusing them of “selective censorship that is harming our national discourse,” and during the Biden administration, Zuckerberg said he was pressured to take down more posts about COVID than he wanted to.

Meta’s abandonment of traditional fact-checking may be cynical, but misinformation is also an intractable problem. Fact-checking assumes that if you can get a trustworthy source to provide better information, you can save people from believing false claims. But people have different ideas of what makes a trustworthy source, and there are times when people want to believe wrong things. How can you stop them? And, the second question that platforms are now asking themselves: How hard should you try?

Community Notes programs—originally invented in 2021 by a team at X, back when it was still called Twitter—are a somewhat perplexing attempt at solving the problem. It seems to rely on a quaint, naive idea of how people behave online: Let’s just talk it out! Reasonable debate will prevail! But, to the credit of social-media platforms, the approach is not as starry-eyed as it seems.

The chief innovation of Community Notes is that the annotations are generated by consensus among people who might otherwise see things differently. Not every note that is written actually appears under a given post; instead, they are assessed using “bridging” algorithms, which are meant to “bridge” divides by accounting for what’s called “diverse positive feedback.” This means that a potential note is valued more highly and is more likely to appear on a post if it is rated “helpful” by a wide array of people who have demonstrated different biases at other times. The basics of this system have quickly become a new industry standard. Shortly after Meta’s announcement about the end of fact-checking, TikTok said that it would be testing its own version of Community Notes, called Footnotes—though unlike Meta and X, TikTok will keep using a formal fact-checking program as well.

These tools are “a good idea and do more good than harm,” Paul Friedl, a researcher at Humboldt University, in Berlin, told me. Friedl co-authored a 2024 paper on decentralized content moderation for Internet Policy Review, which discussed X’s Community Notes among other examples, including Reddit’s forums and old Usenet messaging threads. A major benefit he and his co-author cited was that these programs may help create a “culture of responsibility” by encouraging communities “to reflect, debate, and agree” on the purpose of whatever online space they’re using.

Platforms certainly have good reasons to embrace the model. The first, according to Friedl, is the cost. Rather than employing fact-checkers around the world, these programs require only a simple algorithm. Users do the work for free. The second is that people like them—they often find the context added to posts by fellow users to be helpful and interesting. The third is politics. For the past decade, platforms—and Meta in particular—have been highly reactive to political events, moving from crisis to crisis and angering critics in the process. When Facebook first started flagging fake news, it was perceived as too little, too late by Democrats and reckless censorship by Republicans. It significantly expanded its fact-checking program in 2020 to deal with rampant misinformation (often spread by Trump) about the coronavirus pandemic and that year’s election. From March 1, 2020, to Election Day that year, according to Facebook’s self-reporting, the company displayed fact-checking labels on more than 180 million pieces of content. Again, this was perceived as both too much and not enough. With a notes-based system, platforms can sidestep the hassle of public scrutiny over what is or isn’t fact-checked and why and cleanly remove themselves from drama. They avoid making contentious decisions, Friedl said, which helps in an effort “not to lose cultural capital with any user bases.”

John Stoll, the recently hired head of news at X, told me something similar about Community Notes. The tool is the “best solution” to misinformation, he said, because it takes “a sledgehammer to a black box.” X’s program allows users to download all notes and their voting history in enormous spreadsheets. By making moderation visible and collaborative, instead of secretive and unaccountable, he argued, X has discovered how to do things in “the most equitable, fair, and most pro-free-speech way.” (“Free speech” on X, it should be noted, has also meant platforming white supremacists and other hateful users who were previously banned under Twitter’s old rules.)

[Read: X is a white-supremacist site]

People across the political spectrum do seem to trust notes more than they do standard misinformation flags. That may be because notes feel more organic and tend to be more detailed. In the 2024 paper, Friedl and his co-author wrote that Community Notes give responsibilities “to those most intimately aware of the intricacies of specific online communities.” Those people may also be able to work faster than traditional fact-checkers—X claims that notes usually appear in a matter of hours, while a complicated independent fact-check can take days.

Yet all of these advantages have their limits. Community Notes is really best suited to nitpicking individual instances of people lying or just being wrong. It cannot counter sophisticated, large-scale disinformation campaigns or penalize repeated bad actors (as the old fact-checking regime did). When Twitter’s early version of Community Notes, then called Birdwatch, debuted, the details of the mechanism were made public in a paper that acknowledged another important limitation: The algorithm “needs some cross-partisan agreement to function,” which may, at times, be impossible to find. If there is no consensus, there are no notes.

Musk himself has provided a good case study for this issue. A few Community Notes have vanished from Musk’s posts. It’s possible that he had them removed—at times, he has seemed to resent the power that X has given its users through the program, suggesting that the system is “being gamed” and chiding users for citing “legacy media”—but the disappearances could instead be an algorithmic issue. An influx of either Elon haters or Elon fans could ruin the consensus and the notes’ helpfulness ratings, leading them to disappear. (When I asked about this problem, Stoll told me, “We’re, as a company, 100 percent committed to and in love with Community Notes,” but he did not comment on what had happened to the notes removed from Musk’s posts.)

The early Birdwatch paper also noted that the system might get really, really good at moderating “trivial topics.” That is the tool’s core weakness and its core strength. Notes, because they are written and voted on by people with numerous niche interests and fixations, can appear on anything. While you’ll see them on something classically wrong and dangerous, such as conspiracy theories about Barack Obama’s birth certificate, you’ll also see them on things that are ridiculous and harmless, such as a cute video of a hedgehog. (The caption for a hedgehog video I saw last week suggested that a stumbling hedgehog was being “helped” across a street by a crow; the Community Note clarified that the crow was probably trying to kill it, and the original poster deleted the post.) At times, the disputes can be wildly annoying or pedantic and underscore just how severe a waste of your one life it is to be online at all. I laughed recently at an X post: “People really log on here to get upset at posts and spend their time writing entire community notes that amount to ‘katy perry isn’t an astronaut.’”

[Read: The perfect pop star for a dumb stunt]

The upside, though, is that when anything can be annotated, it feels like less of a big deal or a grand conspiracy when something is. Formal fact-checking programs can feel punitive and draconian, and they give people something to rally against; notes come from peers. This makes receiving one potentially more embarrassing than receiving a traditional fact-check as well; early research has shown that people are likely to delete their misleading posts when they receive Community Notes.

The optimistic take on notes-type systems is that they make use of material that already exists and with which everyone is already acquainted. People already correct each other online all the time: On nearly any TikTok in which someone is saying something obviously wrong, the top comment will be from another person pointing this out. It becomes the top comment because other users “like” it, which bumps it up. I already instinctively look to the comment section whenever I hear something on TikTok and think, That can’t be true, right?

For better or worse, the idea of letting the crowd decide what needs correcting is a throwback to the era of internet forums, where actually culture got its start. But this era of content moderation will not last forever, just as the previous one didn’t. By outright saying that a cultural and political vibe, of sorts, inspired the change, Meta has already suggested as much. We live on the actually internet for now. Whenever the climate shifts—or whenever the heads of the platforms perceive it to shift—we’ll find ourselves someplace else.

This article has been updated to clarify that Meta is ending fact-checking operations only in the United States.