Anthropic researcher says more than 10% chance AI "could kill all humans"

A lead researcher at Anthropic, Evan Hubinger, said he believes there is a more than 10% chance AI “could kill all humans” within the next decade, framing it as a personal view on existential risk. Hubinger’s comments followed the Tuesday resignation of Anthropic researcher Jacob Coxon, who accused Anthropic and OpenAI of acting irresponsibly by racing toward self-improving “superintelligence” without a clear alignment plan. The dispute comes after Anthropic said it has not shared its Claude Mythos 5.1 model with UK and other security bodies outside the United States. The UK’s AI Security Institute says it tested OpenAI’s GPT-6 Astra “only last week” and that risks cross borders.