Researchers challenge AI extinction claims in Nature piece

The trigger was a resignation. Anthropic researcher Jacob Coxon quit on September 8 saying "the people building AI earnestly believe that it could kill us all by the end of the decade," a line the company's own alignment science lead Evan Hubinger amplified with his personal estimate of AI-caused human extinction above 10% within the next decade. Nature's September 22 explainer by Elizabeth Gibney asks whether the science behind that fear actually holds up.

The researchers she talks to answer, mostly, no. Michael Vermeer, a science-and-technology policy researcher at RAND, tells Nature that the doom 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 study he authored looked at three concrete extinction pathways, nuclear, biotechnological, and atmospheric modification, and concluded a full nuclear-driven extinction was infeasible while it could not rule the other two out.

Heidy Khlaaf, chief AI scientist at the AI Now Institute in New York City, is blunter. She calls the extinction scenarios "fear-mongering" and points instead to "AI's low reliability and accuracy rates in critical environments with life-or-death consequences."

Four AI researchers on our tracker shared the piece.