Nature weighs AI extinction warnings after Anthropic exit

Nature's Elizabeth Gibney lines up the extinction warnings now coming out of frontier labs against scientists who say the empirical basis is thin.

The trigger was Jacob Coxon's September 8 resignation from Anthropic. Coxon told the Wall Street Journal he feared the company's AI systems "could spiral out of control and destroy humanity." His colleague Evan Hubinger, who leads Anthropic's alignment-science team, puts the chance of AI-driven human extinction at ">10% within the next decade." Four researchers we track posted the Nature piece the day it ran.

Gibney pairs those claims with sharp pushback. Michael Vermeer at RAND, co-author of a 2025 study on extinction risk from AI, tells her 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." Heidy Khlaaf, chief AI scientist at the AI Now Institute and a former OpenAI safety engineer, calls the framing "fear-mongering" and argues that "Scientific claims require falsifiability precisely to avoid the nature of religious arguments."

On the mechanism, the piece falls back on thought experiments: a paperclip-maximizer superintelligence that renders Earth uninhabitable in pursuit of a narrow goal. Khlaaf wants regulators focused instead on harms Nature names explicitly as present-day: disinformation, bioweapon enabling and false military intelligence.