Ex-Anthropic researcher warns AI race could imperil humanity

A former Anthropic AI researcher, Jacob Coxon, resigns and warns that major labs like Anthropic and OpenAI are racing toward self-improving superintelligence, potentially risking human extinction within the decade; colleagues corroborate the concern and acknowledge the lack of a clear alignment approach, fueling U.S. and EU policy debates as rogue-agent incidents reinforce safety fears.