Worried Anthropic researchers warn that AI ‘could kill all humans’
Covered by 8 sources
Read full post
Anthropic safety lead Evan Hubinger estimates over a 10% chance that AI could kill all humans by 2033, highlighting urgent safety concerns amid rapid AI development. His colleague Jacob Coxon resigned, criticizing the industry's risky race toward uncontrollable superintelligence without clear safety plans.
