The 169-Million-View Resignation

On the evening of September 8th, Jacob Coxon, a 27-year-old who spent three years doing pretraining research at OpenAI and Anthropic, posted: "I resigned from Anthropic today... They are racing straight to self-improving superintelligence and gambling with our lives."

It went from 2,600 followers to 100 million views. In the tracker: nearly 79 million views, then more than 125 million, then 144 million, then 169 million. More than 170 million times viewed.

That is not organic reach. The Wall Street Journal exclusive quoting the resignation went up 18 minutes before his thread. A group chat of about ten people, including the founder of Encode, helped push the post out afterward.

Elon Musk retweeted a post recording the surprisingly high engagement and wrote "Seems like a setup". Skepticism is still amplification.

Coxon is not an outlier. He is the most optimized version of a five-year-old transaction. StartupHub wrote up the resignation when he posted it. The chart below is the rest of the pattern: 20 names, from the 2020 split through this month.

Dario AmodeiDaniela AmodeiGeoffrey HintonYann LeCunFei-Fei LiYoshua BengioIlya SutskeverJan LeikeLeopold Aschenbrenner

WSWilliam Saunders

Daniel KokotajloJohn SchulmanMiles BrundageMira Murati

Mrinank Sharma

ZHZoë HitzigJacob Coxon

JBJoe Benton

JEJosh Engels

Jeff Dean

Portraits are the public photos on Wikipedia and on each person's own profile. Four of the twenty do not have one we could verify, so those seats stay as initials.

The business model

Work at a frontier lab. Quit with a viral warning that AI is out of control. Do the media tour. Convert the fear into capital for a new lab that promises to build the same system, but safely.

Anthropic was the prototype. In December 2020, Dario Amodei, his sister Daniela, and five others left OpenAI over directional differences and founded Anthropic to promote AI safety. Initial funding was $124 million. Last month the company closed a $30 billion Series G at a $380 billion valuation, more than double the $183 billion mark from five months earlier.

Ilya Sutskever left OpenAI on May 14, 2024, and founded Safe Superintelligence Inc. on June 19 with Daniel Gross and Daniel Levy. That was 36 days later. The company raised $1 billion at a $5 billion valuation, then $2 billion at $32 billion.

Mira Murati left as CTO on Sept. 25, 2024. She launched Thinking Machines Lab on Feb. 18, 2025, with about 30 researchers, roughly two-thirds of them ex-OpenAI, and raised $2 billion at a $12 billion valuation.

Yoshua Bengio launched the nonprofit version. LawZero says it will advance research for safe-by-design AI systems. It launched with $30 million in philanthropic funding from Jaan Tallinn, Eric Schmidt, Open Philanthropy, and the Future of Life Institute.

The Superalignment team that was supposed to keep superintelligence aligned no longer exists. Jan Leike, who led it, left in May 2024. Miles Brundage, who led AGI readiness, left that October. Andrea Vallone left a year after. The Mission Alignment team disbanded in February 2026. Bob McGrew and Barret Zoph left with Murati in September 2024. The whistleblower letter that followed the safety exits said: "We joined OpenAI because we wanted to ensure the safety of the incredibly powerful AI systems the company is developing. But we resigned from OpenAI because we lost trust that it would safely, honestly, and responsibly develop its AI systems." Signatories included William Saunders, Daniel Kokotajlo, Carroll Wainwright, Jacob Hilton, and Daniel Ziegler.

John Schulman left for Anthropic to deepen his focus on AI alignment, then left Anthropic five months later. Leopold Aschenbrenner, fired in April 2024, turned the essay into Situational Awareness LP, a hedge fund that reached $45 billion in assets at its peak.

Fei-Fei Li raised $230 million in stealth for World Labs, then $1 billion. Yann LeCun is leaving Meta to found his own world-models startup while Meta rebuilds its AI unit as Superintelligence Labs under Alexandr Wang, former CEO of Scale AI, after putting $14.3 billion into Scale.

The newest variant skips the new lab. Joe Benton headed a safety team at Anthropic. Josh Engels worked at Google DeepMind. Both quit in late August to join METR, the independent evaluator, telling NBC there are "no adults in the room." No raise this time. Just the auditor.

The pundit class that makes it pay

The voices that amplify the warning are holding the cap table.

Former President Barack Obama joined the chorus, saying the technology could be dangerous if it is not managed, and that it is moving very fast in private hands. He urged Democrats to make AI regulation the heart of their agenda. That is the cartel Sam Altman and Dario Amodei have asked for, framed as a way of reining them in.

The billionaires warning the loudest are investors in the labs they say need regulation.

Eric Schmidt is an investor in Anthropic and has talked up the Amazon-Anthropic partnership. In the 2021 Series A the named investors were Jaan Tallinn as lead, Dustin Moskovitz, Eric Schmidt, and James McClave.

Bill Gates and Eric Schmidt were notable investors in Inflection AI, alongside Microsoft. That round raised more than a billion dollars at a $4 billion valuation.

Reid Hoffman, LinkedIn's co-founder, was an early Anthropic investor and has issued the same kind of warning. He also visited Epstein's island in 2014 for MIT Media Lab fundraising.

Jaan Tallinn, Dustin Moskovitz, and the Patrick McGovern Foundation show up in the safety-fund announcements, the Series A, and the open letters about existential risk.

This is why the fear pays. Frontier labs are under commercial pressure to build more powerful systems, and the warnings from Anthropic's CEO and OpenAI's CEO have unsettled investors who already poured billions into the companies that were supposed to benefit. Stocks dip. Then the lab that calls itself the responsible one raises $30 billion at $380 billion. OpenAI, in the same season, announced a $110 billion round, one of the largest private financings on record, at a $730 billion valuation. While Coxon's thread was passing 144 million views, Anthropic was closing the $30 billion round. The same month it removed its flagship safety pledge, the commitment to pause training if capabilities outran the safeguards.

The transaction is a closed loop. An insider supplies the legitimacy. A ten-person group chat and an 18-minute Wall Street Journal lead supply the amplification. Investor-philanthropists validate the danger. The round prices at double the last valuation.

148 million impressions is not 148 million people worried about human extinction. An impression is counted every time a post scrolls past. Next time a grandiose resignation shows up, check the cap table first. The new safe lab is already incorporated.