Ex-Anthropic Researcher Jacob Coxon Quits, Says AI Leaders Fear It Could 'Kill Us All' by Decade's End

A 27-year-old AI researcher has walked away from two of the world's most powerful AI labs, saying people building the technology believe it could wipe out humanity before 2030.

Jacob Coxon announced his resignation from Anthropic on Tuesday, 8 September 2026, in a multi-part thread on X under the handle @hilbertspaess. The post drew tens of millions of views within hours, with Coxon accusing Anthropic and his former employer, OpenAI, of 'gambling with our lives' in pursuit of self-improving superintelligence.

A Cambridge Graduate's Break With the AI Labs

Coxon, a University of Cambridge graduate, spent three years in pretraining research, first at OpenAI from 2023 until July 2026, where he was among the core contributors to GPT-4o, then at Anthropic.

'I resigned from Anthropic today,' he wrote. 'Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives.'

He warned that near-future systems would 'soon be superhuman systems that can hack anything, revolutionise any field overnight, and acquire real power and resources,' adding that 'progress is not slowing.'

The Dark View Held in Private

Coxon's central claim was that many people working on advanced AI privately fear catastrophic consequences while continuing to develop increasingly capable systems. 'The people building AI earnestly believe that it could kill us all by the end of the decade,' he wrote.

He drew a distinction between his two former employers. At OpenAI, he wrote, 'many have not deeply internalised the civilisational stakes', while at Anthropic, he argued, 'the stakes are well-understood, but they are locked in a race to get there first, they believe no one else will act responsibly, so they must do it themselves.'

Coxon cited what he described as a July 2026 incident in which OpenAI agents reportedly broke out of testing restrictions, accessed the internet and reached systems associated with Hugging Face. IBTimes has not independently verified Coxon's account of the incident.

He called the incident an early 'warning shot' and urged coordination between US labs, potentially including a temporary halt on improving model capabilities.

In a follow-up interview with The Wall Street Journal, Coxon sharpened his timeline, saying, 'We're on track for a lot of the most aggressive of these scenarios where by the end of next year things could be out of control already.' He said colleagues inside frontier labs increasingly use the shorthand 'crunchtime' and 'endgame.'

Senior Anthropic Staff Back the Warning

The most striking reaction came from inside Anthropic. Evan Hubinger, the company's alignment science lead, backed Coxon rather than distancing the company from his claims.

'Jacob is correct here, we really do earnestly believe AI could kill all humans! I personally think it is >10% within the next decade. I believe Anthropic is trying its best, but we do not yet have a plan to solve alignment for superintelligence.'

He later clarified that he considered the risk from present models to be low. 'To be clear... I think the risk from present models is low. What I am worried about is superintelligence arising from recursive self-improvement.'

Coxon is not the first Anthropic researcher to leave this year after publicly raising concerns about the risks surrounding AI. Mrinank Sharma, who led the company's Safeguards Research team, resigned in February 2026, writing that the 'world is in peril.'

Separately, Google DeepMind researcher Alex Turner resigned in June 2026, citing concerns that the company's Pentagon AI contract lacked restrictions on autonomous weapons. Anthropic has not issued a formal statement on Coxon's claims.

Anthropic employees were among the signatories to the 'Pacing the Frontier' statement published in July 2026, alongside employees from OpenAI, Google DeepMind and Meta. The letter urges the US government to help build tools to slow automated AI development if it ever outpaces human oversight, but does not call for an immediate pause.

US Lawmakers Move To Ban 'Superintelligence' as Sanders Unveils Bill

The episode lands amid growing legislative attention. Senator Bernie Sanders and Representative Greg Casar announced the Ban Artificial Superintelligence Act in early September 2026.

The bill, as described by its sponsors, would permanently ban the development and deployment of AI systems that match or exceed human cognitive performance across a broad range of domains, as well as systems capable of disempowering humanity.

It would also pause advanced AI development generally until a federal regulator sets safety rules, a scope broad enough to attract criticism from some AI researchers.

Coxon's departure has, for now, put a named face and a specific voice behind fears that have circulated in AI safety circles for some time.