Anthropic researcher says AI has more than 10% chance of 'killing all humans' after colleague quits

An Anthropic safety researcher, Evan Hubinger, stated on Tuesday that he personally believes there is a greater than 10% chance AI could kill all humans within the next decade. This comment came hours after another Anthropic researcher, Jacob Coxon, resigned, accusing AI labs like Anthropic and OpenAI of "gambling with our lives" by racing toward self-improving superintelligence. Coxon warned that such systems could soon become superhuman, hack anything, and acquire real power, and noted that people building AI earnestly believe it could be an existential threat by the end of the decade. Hubinger agreed with Coxon's assessment and added that Anthropic currently has no clear plan to solve the alignment problem for superintelligence, though he clarified the risk from current models is low. The concerns highlight a growing tension within leading AI labs between advancing powerful technology and ensuring its safety, as both Anthropic and OpenAI continue to raise significant capital.

(Source：CNBC)