Anthropic alignment lead pegs AI extinction odds above 10%

Anthropic's alignment science leader Evan Hubinger has put the odds of AI driving humanity extinct at more than 10% within the next decade. Nature reports the statement accumulated over 100 million views in 24 hours.

The number arrived weeks after Jacob Coxon resigned from Anthropic and told the Wall Street Journal that AI systems "could spiral out of control and destroy humanity." CEO Dario Amodei answered with a call to slow, not halt, frontier development. OpenAI's Sam Altman and xAI's Elon Musk endorsed that posture.

The scientists Nature quotes are less convinced. Michael Vermeer at RAND says such predictions "involves so many untestable claims that you just end up with a conversation that is really more like faith than something scientific." RAND's own 2025 analysis found complete extinction via nuclear weapons unfeasible, though biotechnology and atmospheric modification scenarios "could not be ruled out."

Heidy Khlaaf, chief AI scientist at the AI Now Institute, calls the extinction framing "fear-mongering" and argues that "AI's low reliability and accuracy rates in critical environments" are the clearer harm. The piece also reproduces the stock philosophical example verbatim: a hypothetical superintelligence that "aims to manufacture as many paper clips as possible could end up making Earth uninhabitable." A separate 2027 forecast from the AI Futures project imagines a scenario in which "an AI unleashes a biological weapon to kill humans off."

Nature does not duck the commercial backdrop. The article says Anthropic's safety positioning may reflect genuine concern alongside potential liability exposure and upcoming IPO considerations. It lands as nearly 1,400 AI-sector employees have signed an open letter demanding a slowdown and US legislators push to ban "artificial superintelligence" outright. Four of the researchers we track in our Who's Who shared the piece the day it ran.