Nature finds thin science behind Anthropic extinction alarm

When Anthropic researcher Jacob Coxon resigned in September with a warning that superintelligent AI could cause human extinction, his post hit 100 million views in 24 hours. In Nature, Elizabeth Gibney examines the scientific case under the alarm and finds independent researchers unconvinced.

In a Slack message to colleagues, Coxon wrote that "the people building AI earnestly believe that it could kill us all by the end of the decade." Evan Hubinger, who leads alignment science at Anthropic, publicly backed him, estimating the risk of extinction at ">10% within the next decade".

Researchers outside the frontier labs tell Nature those probabilities rest on claims nobody knows how to test. Michael Vermeer at RAND said the forecasts "involve so many untestable claims that you just end up with a conversation that is really more like faith than something scientific or empirical." A 2025 RAND report Vermeer co-authored examined specific extinction pathways. Complete extinction by nuclear weapons was judged infeasible. Biotechnology and large-scale atmospheric modification could not be ruled out, but either would require considerable ability to physically interact with the world and would take time, leaving room for human intervention.

Heidy Khlaaf, chief AI scientist at the AI Now Institute and a former OpenAI safety engineer, is blunter. She calls the extinction framing "fear-mongering" and tells Nature regulators should focus instead on disinformation, bioweapon creation and false military intelligence, including a recent incident in which AI-generated intelligence nearly triggered an unauthorized boarding of a Chinese vessel. Her policy line: "AI labs cannot continue to define the course of AI governance."

Four researchers in our tracked directory shared the piece.