“I resigned from Anthropic today,” wrote the previously obscure Anthropic researcher Jacob Coxon Tuesday in The Tweet That Shook The World. Neither OpenAI nor Anthropic, his former employers, is “acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives.”
I resigned from Anthropic today. I spent the last three years doing pretraining research at both OpenAI and Anthropic. Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives. More thoughts below.
— Jacob Coxon (@hilbertspaess) September 9, 2026
Within 24 hours, Coxon’s Tweet had more than 100 million views, amplified by politicians, mostly prominent Democrats, who sounded the claxon. “It’s becoming more clear the threat AI poses to humanity, so I’m calling for immediate action from the industry and Washington,” wrote Illinois Governor J.B. Pritzker, who never saw an alarm bell he couldn’t raise. Connecticut Senator Chris Murphy invoked “AI companies in a blind race to build a death machine first.” Bernie Sanders once again is asking you to “ban superintelligence” (a serious threat to his very normal intelligence) and “pause AI development.”
Coxon went on a media blitz, telling the Wall Street Journal it’s “crunch time” for humanity. Wired Magazine magnified his claims, quoting Coxon as saying that crunch time was actually “the next two years.” Another Anthropic researcher, Evan Hubinger, publicly said he puts the probability of AI killing all humans within the next decade above 10%, which seems a little high.
Axios literally ran a story headlined “Anthropic insiders warn AI could kill all humans,” calling the warnings “chilling.” Ars Technica used “Self-improving AI could ‘kill us all,’” while TechCrunch wrote that uncontrolled AI development could “end up killing us all.”
As the Vince McMahon meme goes, triggering intensified. Business Insider published “Researchers fear there’s a chance AI could kill us all. Here’s how experts say that might play out,” walking readers through cyberattacks, infrastructure collapse, biological weapons, and autonomous systems taking power. The New York Post went with “Experts warn AI could kill us all — this is how an apocalyptic doomsday scenario could play out,” extending the possibilities to electrical grids, water systems, bioweapons, nuclear catastrophe, and the cheerful conclusion that even hiding in a bunker might not save you.
The Guardian similarly framed the question as whether AI might “kill us all,” assembling Coxon, Hubinger, Stuart Russell, Geoffrey Hinton, Toby Ord, and other existential-risk voices alongside harrumphing demands for international regulation and restrictions on superintelligence. CBS Boston found an academic to tell viewers that Coxon’s warning needed to be taken seriously. Even People ran the ridiculous tabloid headline: “Employee at Top AI Company Quits with Public Warning the Tech Could ‘Kill Us All,’ and Then Former Co-Worker Agrees.”
Coxon is a real person who had a real career in AI. And his concerns seem at least somewhat genuine. But the full-court press doesn’t. The media first reported a resignation containing a speculative extinction forecast as an alarming testimony from inside the AI labs, then rapidly converted it into a media event about how and when artificial intelligence might actually wipe out the human race. Stop right there, as Meat Loaf once sang in more analog days, before we go any farther.
The money trail behind the AI-doom movement runs through a remarkably small circle of AI-safety philanthropy. Skype co-founder Jaan Tallinn, an early Anthropic investor, is also a major funder of the advocacy ecosystem surrounding AI existential risk: his Survival and Flourishing Fund directed more than $2 million to the AI Futures Project in 2025, along with $516,000 to Encode AI and $1.635 million to the AI Policy Institute. Dustin Moskovitz, another investor in Anthropic’s original funding round, and his wife Cari Tuna are the principal funders of Coefficient Giving, formerly Open Philanthropy, another major financial engine of the AI-safety movement.
None of this proves anyone orchestrated Coxon’s resignation. But it does complicate the image of an entirely spontaneous uprising of terrified independent experts. The AI-doom movement has some extremely wealthy patrons and professional advocates working from inside well-funded institutions. And its message has already found eager allies in Washington, particularly among Democrats but among Republicans as well, in an election season where public opinion about AI is growing increasingly hostile.
Coxon “spontaneously” revealed his judgment about where current progress might lead and reported what colleagues supposedly believe privately. Yet much of the coverage treats the credentials of the people intensely expressing the fear almost as evidence for the underlying proposition, when in fact they’re just credentials.
Of course there are issues involved with the rapid march of AI into most human activity. But can we turn down the moral panic dial, please, and maybe think about who these Cassandras actually are and why we might be hearing from them now? “The apocalypse is imminent” isn’t an acceptable or a rational answer.
As of today, humanity hasn’t fallen. Though I did just see a sunglasses-clad man driving down my street on a motorcycle while brandishing a shotgun. Maybe I shouldn’t have used ChatGPT to generate that meme last night after all.
***
Neal Pollack is the author of 12 semi-bestselling books of fiction and nonfiction and is a three-time Jeopardy! champion.
.png)
.png)
