Anthropic Researcher Quits, Warning AI Labs Are 'Gambling With Our Lives

An AI safety researcher who left OpenAI to join Anthropic has resigned from the company and announced he is leaving the AI industry entirely, accusing both US labs of "gambling with our lives" in the race to build models capable of recursive self-improvement. Anthropic safety executive Evan Hubinger publicly backed the researcher on X, writing that "We really do earnestly believe AI could kill all humans!" and putting the risk of human extinction at more than 10 percent over the next decade. The resignation has been covered by CNBC, The Guardian, NDTV, Independent.ie and commentator Derek Thompson, who frames it as fuel for global calls for AI regulation. Coverage notes that the departure comes amid a broader wave of "extinction" warnings from current and former OpenAI and Anthropic staff urging a slowdown in frontier model development. The outlets also revisit past public statements on risk by OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei, contrasting those warnings with the continued pace of commercial AI releases. No new regulatory action has been announced, and neither company has issued a formal response in the material available.

Global Impact

Technologically, the episode sharpens the split between labs racing toward recursive self-improvement and staff who believe the risk of catastrophic loss of control exceeds 10 percent, which could slow hiring and release timelines at frontier labs. Politically, it hands regulators in Washington, Brussels and London fresh testimony-grade material to justify binding safety evaluations, licensing or compute thresholds, with the EU AI Act's implementation phase as the nearest concrete venue.