Jacob Coxon walked away from Anthropic on Sept. 9. He posted that the lab, along with OpenAI, races toward self-improving superintelligence while gambling with human lives1. Four Anthropic scientists answered within a day. Every one of them stayed on the payroll. Evan Hubinger, who leads alignment science at the company, wrote that he personally holds the odds of AI causing human extinction above 10 percent within the next decade2. Samuel Marks, Anna Wang, and Ethan Perez each added their names to the same admission. One departure became a five-person public reckoning. It ran in real time on a platform built more for arguments than confessions.
The question outlasts the news cycle. When the people closest to a technology say in public that it could end humanity, does staying inside change anything?
A Resignation as Argument
Jacob Coxon spent three years on pretraining research, first at OpenAI, then at Anthropic1. He announced his exit on Sept. 9. The post read as an indictment, and barely as a goodbye. His central claim lands as an accusation: "They are racing straight to self-improving superintelligence and gambling with our lives". Read plainly, the sentence names two defendants and one shared crime. Its subtext runs deeper in the possessive. "Our lives" folds Coxon, and every reader, into stakeholder status inside a decision made in buildings he had already left.
Quitting became his argument more than his exit. By Sept. 10 the post had drawn nearly 76 million views18.
The Colleagues Who Chose to Stay
Evan Hubinger broke the silence first. He posted within hours that "we really do earnestly believe AI could kill all humans! I personally think it is >10% within the next decade"2. The exclamation point matters more than the percentage. A researcher using punctuation built for enthusiasm to deliver a statement about mass death signals resigned candor. Bluntness may be the only register left.
Samuel Marks, who leads scalable oversight at the company, followed with a starker frame: "AI developers believe their technology could cause human extinction"1. He works on safety research at Anthropic because he hopes it will reduce the chance of what he called "extinction-level bad outcomes". That sentence reads as a mission statement and a confession at once. Ethan Perez, who leads the alignment team, closed the loop with the shortest entry: "100% agree with him that AI poses serious risks to society, and I'm glad he's speaking out!". Brevity carries its own signal. When the alignment lead needs one sentence to concur with an extinction estimate, the concurrence weighs more than any elaboration.
The Missing Plan
Hubinger described Anthropic as trying its best2. The company works from behind on alignment for superintelligence, still hunting a plan that closes the distance. Anna Wang, a member of Anthropic's technical staff who previously worked at Google DeepMind, went further1. A workable plan for controlling recursively self-improving AI is still a work in progress, she said. That holds across the entire field. Wang called staying inside a lab the harder route. In her judgment it is also the more effective one, and she called the calculation anything but easy.
Marks supplied the clearest account of why researchers who believe the stakes run lethal keep showing up to work1. Commercial incentive competes against one belief. Rivals would build faster and less carefully if the cautious labs stepped back. Together, the four statements sketch one organization. It races to close the distance between a published safety roadmap and the pace of its own releases2.
A Familiar Pattern, A New Twist
Anthropic's week echoes a scene the industry already lived through. Jan Leike resigned from OpenAI's superalignment team in May 2024, writing that "safety culture and processes have taken a backseat to shiny products"6. Leike left. Hubinger, Marks, Wang, and Perez stayed, and that distinction reframes the episode. Departure once read as the loudest tool available to a worried researcher. Public agreement from people who kept their badges reads louder now. It carries the credibility of standing close to the systems they call dangerous.
Dario Amodei built Anthropic on the premise that safety work happens more effectively inside a frontier lab than outside one1. This week four of his researchers tested that premise in public, on the record, with their names attached2. The premise now carries a price tag: four researchers on payroll, each naming a number, each drawing a salary from the company whose trajectory worries them most.
The Competition That Keeps the Race Running
Marks named the mechanism directly1. AI developers keep building because they compete against rivals they judge less careful. Caution becomes a competitive liability every lab claims to regret, while repair waits for whichever rival moves first. Chris Lehane, OpenAI's chief global affairs officer, published his own appeal to Congress the same week7. He argued that "the prospect of AI-accelerated AI development demands more than voluntary commitments" and that the country needs binding, capability-based rules. Two labs, in the same seven days, reached versions of one diagnosis. Internal caution alone struggles to outrun a race every competitor keeps running. Regulation slows every runner at once.
What Governance Can Actually Do
Anthropic operates under a public Responsible Scaling Policy1. It ties model releases to capability thresholds. Anthropic built it to keep pace with warnings of the kind just issued2. Whether that framework holds under the pressure Coxon named requires cooperation past Anthropic alone, reaching OpenAI, Google DeepMind, and every other lab chasing the same frontier. Governance built inside one company answers only part of the question Marks raised about incentives across an industry.
Employees speaking in public, ahead of filing internal memos, work as a pressure campaign aimed outward at regulators and rival labs, and inward at Anthropic's own leadership12. Pressure of that kind carries a cost. Every public estimate invites scrutiny of Anthropic's own hiring, retention, and product timelines. The standard its researchers just set now applies to the company itself.
Jensen Huang took the opposing position, and from outside any alignment team entirely. Nvidia's co-founder and chief executive answered Coxon's post directly in a Sept. 20 interview with CBS News. "2030 is not going to be the end of the world. There is 0% chance that's going to be the end of the world," Huang said. He called estimates like Hubinger's "not grounded in science"9. The dismissal reframes the question Marks raised about competition: Huang runs the company that sells the chips every lab in this piece races to buy, and his stake in a faster pace comes with its own set of interests.
Coxon's Sept. 9 resignation post neared 76 million views by Sept. 10, and Nvidia's chief answered it on Sept. 20
Dated events after Jacob Coxon left Anthropic, Sept. 9 to Sept. 20, 2026
Each pin is one dated event on a rail spaced by real time: magenta marks the resignation and its reach, and blue marks the answer from Nvidia's chief executive.
Sources: HuffPost [1]; Quartz [8]; CBS News [9]. Chart by Ryan Elliott Dennis.
The numbers behind this chart
| Date | Event |
|---|---|
| Sept. 9, 2026 | Coxon leaves Anthropic. He posted that the lab races toward self-improving superintelligence |
| Sept. 10, 2026 | Post nears 76M views. Post nears 76 million views. Within a day of publication |
| Sept. 20, 2026 | Huang answers on CBS. Huang answers on CBS News. Jensen Huang called a 2030 end of the world a 0% chance |
By the numbers
- Greater than 10 percent: Evan Hubinger's personal estimate of the odds AI causes human extinction within the next decade2.
- Three years: length of Jacob Coxon's pretraining research career across OpenAI and Anthropic before his Sept. 9 resignation1.
- Four: named Anthropic researchers, beyond Coxon, who spoke publicly on the extinction-risk question within a single week12.
- May 2024: month OpenAI's Jan Leike resigned over safety culture concerns, a comparable public moment6.
- Within hours: span between Coxon's resignation post and Hubinger's public reply12.
- 76 million: views Coxon's resignation post drew within a day of publication8.
What to watch
Anthropic's next model release tests the Responsible Scaling Policy1. Four researchers put a number on the stakes in public, and the policy has to answer it2. Congress adjourns in December, and Lehane's parallel push for binding rules gives regulators a concrete deadline7. Watch whether researchers at other labs follow Hubinger and attach a number to their own private estimates, a move that would turn one company's reckoning into an industry standard for disclosure.
Sources
- HuffPost, "More Anthropic Employees Sound The Alarm About AI Safety Risks," HuffPost, Sept. 9, 2026, https://www.huffpost.com/entry/anthropic-ai-risks_n_6aa1d186e4b086ecc55fc61d
- Officechai, "Anthropic Alignment Science Lead Evan Hubinger Says There's A More Than 10% Chance AI Could Kill All Humans Within Next Decade," Officechai, Sept. 9, 2026, https://officechai.com/ai/anthropic-alignment-science-lead-evan-hubinger-says-theres-a-more-than-10-chance-ai-could-kill-all-humans-within-next-decade/
- CBS News, "Anthropic Researcher Says More Than 10% Chance AI 'Could Kill All Humans,'" CBS News, Sept. 9, 2026, http://www.cbsnews.com/news/ai-kill-humans-anthropic-researcher-more-than-ten-percent-chance/
- The Hill, "Anthropic Researchers Warn AI Could Kill Humans by the End of the Decade," The Hill, Sept. 10, 2026, https://thehill.com/policy/technology/6078907-anthropic-researchers-warn-ai-extinction/
- Siladitya Ray, "Anthropic Alignment Lead Issues Warning About AI Killing Humans As Researcher Resigns," Forbes, Sept. 9, 2026, https://www.forbes.com/sites/siladityaray/2026/09/09/anthropic-alignment-lead-warns-ai-could-kill-all-humans-as-researcher-quits/
- CNN Business, "More OpenAI Drama: Exec Quits Over Concerns About Focus on Profit Over Safety," CNN, May 17, 2024, https://www.cnn.com/2024/05/17/tech/openai-exec-exits-safety-concerns
- KFGO, "OpenAI Pushes for Mandatory National AI Safety Requirements," KFGO, Sept. 9, 2026, https://kfgo.com/2026/09/09/openai-pushes-for-mandatory-national-ai-safety-requirements/
- Quartz, "Jacob Coxon Quits Anthropic Over Self-Improving AI Safety Fears," Quartz, Sept. 9, 2026, https://qz.com/anthropic-researcher-quits-self-improving-ai-safety-090926
- Jo Ling Kent, Emily Pandise and Aimee Picchi, "Nvidia's Jensen Huang Rejects AI Extinction Warnings As 'Doomsday Narratives,'" CBS News, Sept. 20, 2026, https://www.cbsnews.com/news/jensen-huang-nvidia-rejects-ai-extinction-warnings/