Anthropic Researcher Warns AI Could Kill All Humans

Anthropic’s Alignment Science Lead Evan Hubinger publicly stated a greater than 10% chance AI could kill all humans within a decade, days after a colleague resigned over safety concerns.

One of the most alarming warnings ever issued from inside a leading AI laboratory has gone viral, and it came not from a critic on the outside but from a senior safety researcher at Anthropic itself. Evan Hubinger, Anthropic’s Alignment Science Lead, posted on X that he personally believes there is a greater than 10% chance AI “could kill all humans” within the next decade, a statement that has since been viewed more than 10 million times and triggered a global debate about who is responsible for preventing that outcome.

Hubinger’s post was a direct response to Jacob Coxon, an AI researcher who had just quit Anthropic and previously worked at OpenAI, and who wrote in a resignation thread that “the people building AI earnestly believe that it could kill us all by the end of the decade,” adding that “neither company is acting responsibly.”

Hubinger confirmed the sentiment: “Jacob is correct here; we really do earnestly believe AI could kill all humans! I personally think it is greater than 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 and are not clearly on track to.”

Hubinger was careful to qualify the source of the risk. Current models, he said, posed a “low” threat. What worried him was the prospect of superintelligence arising from recursive self-improvement, AI systems that can autonomously upgrade their own capabilities, potentially reaching a point where human oversight becomes impossible or irrelevant.

The timing of the disclosures is significant. The remarks followed a Financial Times report that Anthropic had withheld its latest model from the UK’s AI Safety Institute (AISI), one of the world’s most prominent bodies for independent AI risk assessment.

A Cabinet Office spokesperson declined to confirm whether the model had been withheld, saying only that the government “continues to collaborate closely with industry partners, including Anthropic, to make models safer.”

Dame Wendy Hall, a computer scientist who advises the UN on AI, told BBC Radio 4 she was “shocked” by the posts, though she acknowledged some of it could be “PR and marketing” as Anthropic and OpenAI race toward stock market listings.

She added: “Why would someone want to say that? I would plead with investors not to invest in this company if that is their value system.”

Neil Lawrence, Professor of Machine Learning at the University of Cambridge, told the BBC the report was credible and pointed to a broader shift in the US posture on AI oversight.

The debate is no longer about whether AI poses a risk, but about how large that risk is and whether the companies building it are moving fast enough to address it. For now, one of Anthropic’s own researchers has answered that question with a number: greater than 10%.