Researchers warn AI ‘could kill us all’ by the end of the decade

A safety researcher at the AI company Anthropic has warned the technology is advancing so quickly that there’s a chance it could “kill all humans” in the next decade.

Evan Hubinger said although the current risk was “low”, he was “worried” AI systems could improve themselves to the point where they posed an existential risk.

He was responding to a post on X (formerly Twitter) by another researcher, Jacob Coxon, who said he had quit the company amid concerns it is “gambling with our lives”.

Coxon tweeted: “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.”

Artificial superintelligence is when the technology exceeds human intelligence and “self-improving” refers to its ability to produce more capable versions of itself.

Coxon said AI developers “earnestly believe that it could kill us all by the end of the decade”. He said no other human activity posed the same level of danger.

In his post, Hubinger said “we really do earnestly believe” AI poses a species-ending risk to humans.

“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.”

A third Anthropic employee, Samuel Marks, tweeted: “AI developers believe their technology could cause human extinction (or similarly bad outcomes). This could happen in the next few years. In general, the more senior the employee, the more concerned they are.”

He said many AI developers “desperately want to slow down” the pace with which the technology is advancing.

In a separate development, the Financial Times reported that Anthropic withheld its latest model from the from the UK's AI Safety Institute (AISI).

Anthropic has declined to comment on the posts by its employees.