What should we make of the AI doomsday campaign?

Jacob Coxon, a 27-year-old researcher, publicly resigned from his AI training role at Anthropic on Tuesday, accusing Anthropic and OpenAI of "racing straight to self-improving superintelligence and gambling with our lives," and on Saturday Anthropic founder Dario Amodei published an essay calling for a slowdown of AI development that OpenAI chief Sam Altman, xAI leader Elon Musk, and Google DeepMind co-founder Demis Hassabis have since backed to varying degrees. Amodei's essay argues the US government should waive antitrust regulation so his company and its competitors can coordinate, drawing accusations of "regulatory capture" online as both companies pursue historic IPOs, though OpenAI is reportedly delaying its offering amid AI safety fears.

While the normies were resting, we mainlined tech discourse all weekend to bring you the latest trends, rumors, fights, and innovations from the most manic corners of the internet. This week: The public debates what to make of the AI doomsday campaign, Mayor Lurie’s statement on SF’s rent crisis is outshined by royalty-free background music, and Apple’s future rests in anatomically-impossible hands. This is Manic Monday. Doomsday, political conspiracy, or a secret third thing? Any given weekend, when I ride along the information superhighway in search of good tidbits for this column, there are generally a handful of distinct and equally manic conversations to choose from. This weekend was different. I couldn’t open my phone without immediately drowning in a flood of AI doomsday scenarios and panicked debates over regulation. It started last Tuesday, when a 27-year-old researcher named Jacob Coxon publicly resigned from his role training AI at Anthropic, taking to X https://x.com/hilbertspaess/status/2097476196791709843 to accuse his employer and OpenAI his former employer of “racing straight to self-improving superintelligence and gambling with our lives.” The story quickly went viral; other AI researchers echoed https://x.com/a nnawang/status/2097720574500102615 his https://x.com/saprmarks/status/2097570226804011302 concerns https://x.com/EthanJPerez/status/2097861257714172270 on social media, while Coxon was ushered onto major news channels including CNN https://www.cnn.com/2026/09/09/tech/ai-anthropic-safety and BBC https://www.bbc.com/news/articles/c1kx0gyje9wo for interviews, telling the latter that those closest to the technology believe it poses “a possibility of human extinction.” Then on Saturday, the conversation reached a fever pitch when Anthropic founder Dario Amodei published an essay https://darioamodei.com/post/we-must-pace-the-frontier calling for a slowdown of AI development, a proposal that his rivals, OpenAI chief Sam Altman https://x.com/sama/status/2098811563415150910 , XAI leader Elon Musk https://x.com/elonmusk/status/2098789109980332057 , and Google DeepMind co-founder Demis Hassabis https://x.com/demishassabis/status/2098909516582490602 , have all since pledged their support to varying degrees. Why is this happening now? Coxon is not the first AI researcher to sound the alarm. AI people have been debating P doom , a made-up measurement for the probability that rogue bots will kill us all, for years. While I do believe that AI insiders may be genuinely frightened at how fast the technology is advancing, especially after the incident last month where OpenAI agents autonomously hacked into an open-source code library https://www.reuters.com/legal/litigation/openai-agents-attacked-software-service-rubygems-before-hugging-face-incident-2026-09-11/ called Hugging Face, there are other factors at play. In Amodei’s essay, he argues the US government should waive antitrust regulation so his company and its competitors can coordinate — in other words, form a legal cartel — for the sake of humanity. Online, the phrase “regulatory capture” is starting to enter the vernacular in both conservative https://x.com/Jason/status/2098819632178119081 and progressive https://www.instagram.com/p/DdH71ocDrxi/?img index=6 circles. Conspiracy theories about coordinated media campaigns https://x.com/parkerthayer/status/2097759699626328575?s=10 and Chinese psyops https://x.com/spencerpratt/status/2098121678534377961 abound. The generous take is that insiders really have seen something scary in their labs, and are now balancing their moral and commercial interests in good faith. But with both companies pursuing historic IPOs soon although OpenAI seems to be delaying https://www.reuters.com/legal/litigation/openai-ipo-will-not-happen-2026-amid-ai-safety-fears-altman-says-2026-09-12/ , the cynical take is that they want more control over price-setting and profits to make up for embarrassing financials and a bubble threatening to pop. The truth, as it always is, is probably somewhere in between. The only thing I’m certain of is that AI regulation will be a major part of both parties’ platforms come midterms. — Cydney Hayes In other news, Elizabeth Holmes goes wine mom mode in a selfie posted from prison Besides the AI doomsday spiral, this was quite literally the only other conversation happening on my X feed this weekend. The account of convicted fraudster Elizabeth Holmes posted a selfie https://x.com/ElizabethHolmes/status/2099324752863379523 of her blowing a kiss to the camera. I say “the account of” because inmates are not allowed cell phones at Federal Prison Camp Bryan, where Holmes is currently incarcerated. Her blush is poppin’, her camera is potato-quality, and her finger is adorned with a snake ring evidently https://x.com/ElizabethHolmes/status/2099331765198524721 purchased from Taylor Swift’s line of Reputation merch. She captioned the photo, “The answer to every challenge lies in your heart.” Very Instagram, circa 2013. Did Holmes instruct someone to post this while she’s in prison? How does it fit with her post-conviction PR strategy? Is she a Swiftie? These are questions perhaps no one but Nathan Fielder https://www.youtube.com/watch?v=GGJSRFWALTI will ever know the answer to. — C.H. Finger s on the pulse Apple debuted its folding iPhone Duo https://www.apple.com/newsroom/2026/09/apple-unveils-iphone-duo/ late last week, and it looks like a fascinating evolution of tech with the company’s typical aesthetic polish. A dose of controversy came, however, when eagle-eyed observers realized something odd in Apple’s marketing: An image that depicts an open Duo being held in a hand with impossibly long fingers. https://x.com/Polymarket/status/2097777885893292337?s=20 I adore deeply stupid online “controversies” that effectively boil down to conspiratorial stoner thoughts, and the theory that Apple would’ve manipulated an image to make it look like its folded-open Duo fits neatly in a totally human, not elongated hand is peak internet. Naturally, shitpost X accounts are making parodies $299 iFingers, anyone? https://x.com/the fauxy/status/2097953410301718994?s=20 while other techies are earnestly chiming in with concern. See: prolific founder Martin Varsavsky https://en.wikipedia.org/wiki/Mart%C3%ADn Varsavsky questioning why a phone demands “ fingers like a concert pianist https://x.com/martinvars/status/2097949605744804305 ” to hold. Is the whole thing completely fabricated based on a simple optical illusion https://x.com/stephenrobles/status/2098237472618487984?s=20 that makes the Duo hand look enlarged? Possibly. But should we drive down to Cupertino to protest Apple and demand the truth? Probably — Eddie Kim The sound of silence For as much as Mayor Daniel Lurie is celebrated as a social media whiz https://sf.gazetteer.co/our-mayor-is-posting-through-it , a kind of Mamdani of the West, if you will, New York City’s mayor would never release a short-form video as comically bad as the one our mayor posted last week. In a now-deleted 15-second TikTok uploaded Friday afternoon https://lede-admin.sf.gazetteer.co/wp-content/uploads/sites/57/2026/09/v15044gf0000dai3jifog65jmqfsbulg.mp4 , Lurie acknowledges the city’s rent crisis at the hands of an AI money influx and the landlords that want to capitalize on it. It is only the second video he’s shared where he explicitly discusses this issue. “San Francisco is in a rent emergency,” Lurie says, before he is interrupted by a sunny, royalty-free guitar riff that drowns his voice out altogether. You can read what he says because the video has been subtitled: Lurie’s introduced a slate of legislation to address the rent emergency, and he is excited to work with the Board of Supervisors to get things rolling. The original video was quickly deleted, but Gazetteer SF saved the video for posterity. He uploaded a fixed version https://www.instagram.com/reel/DdIKmxXyGgA/?stkn=cXZ4ZHN0bWZ2N3hk , without any backing track, to Instagram; this version has yet to make its way to TikTok. Lurie wasn’t saying much of substance anyway — no rent freezes or hard bans on rent hikes, to be clear — but you can’t even hear him as he said it It’s like the Boyz n the Hood saxophone meme https://knowyourmeme.com/memes/hood-saxophone-90s-movie-saxophones , where the situation gets more dire as the music plays on, except this is an official video from the official mayor’s account. For all of the glossy videos he’s made about meeting Pikachu https://sf.gazetteer.co/i-for-one-welcome-our-new-pokemon-overlords , or a buzzy restaurant opening, or the new Uniqlo opening in downtown, you’d think that this very important video would have been made a lot less hastily. Or maybe this was just the first time Lurie’s making the TikTok himself, instead of one of his many Zoomer handlers. — Joshua Bote AI’s odd couple A majority of Americans now feel more concern than optimism about the technology, per an August poll https://www.pewresearch.org/short-reads/2026/08/18/young-adults-in-the-us-are-increasingly-wary-of-ai-concerned-it-will-take-jobs/ from the Pew Research Center. No wonder, then, that anti-AI sentiment is bringing two very strange political bedfellows together: Sen. Bernie Sanders and former Trump advisor/current sentient cane toad https://static01.nyt.com/images/2022/03/29/science/15tb-cannibal-toads1/15tb-cannibal-toads1-mediumSquareAt3X.jpg Steve Bannon. Sanders and Bannon form the centerpiece of the Future of Life Institute’s https://futureoflife.org/ “Pro-Human Assembly” in DC on Tuesday, which is part of the think-tank’s efforts to “protect human agency” and keep tech corporations accountable. Sanders has spent much of the year working to pause “superintelligent AI," https://futurism.com/artificial-intelligence/bernie-sanders-banning-superintelligent-ai-imprisoning-developers-20-years but to his credit, Bannon’s been ringing the alarm https://thehill.com/homenews/administration/5154889-steve-bannon-oligarchs-republicans/ on oligarchs and Big Tech for a while, sounding kind of like a Marxist with all his talk about late-stage capitalism putting “ the nation and mankind in jeopardy. https://www.axios.com/2026/09/14/coxon-bernie-bannon-unite-human-ai ” Bannon is still gross and evil https://www.vanityfair.com/news/story/steve-bannon-jeffrey-epstein-files-trump-maga , mind you, but anti-AI populism is an intriguing next step for any far-right architect, especially if someone as beloved as Bernie is co-signing Bannon’s takes. Meanwhile, Bannon’s ex-flame https://www.bbc.com/news/world-us-canada-42636667 Donald Trump is out here declaring https://x.com/RapidResponse47/status/2099498628410646945?s=20 that the only AI guardrail we need is a “High IQ ” president oh and claiming Anthropic’s Dario Amodei is pretending to be a “perfect little angel.” Frankly, I think Amodei might be a big ol’ hypocrite https://medium.com/predict/forget-sam-altman-dario-amodei-is-far-more-dangerous-heres-why-51f6d9d72ac8 , but I’m not too worried about Trump: He clearly doesn’t know how to use AI https://x.com/Acyn/status/2099284092626129214?s=20 , let alone how to think about it. — E.K. The week ahead : Elizabeth Holmes starts live-streaming her DIY jewelry business.