Anthropic researcher: Over 10% chance AI could kill all humans

A senior safety researcher at AI firm Anthropic has stated there is a greater than 10 percent chance that artificial intelligence "could kill all humans" by the end of this decade. The warning came just hours after a colleague resigned, citing concerns that AI labs and their rivals are carelessly racing to build "superhuman systems" they cannot control.

Jacob Coxon, a researcher who previously trained AI systems at Anthropic and OpenAI, announced his departure on X. He accused both AI companies of "racing straight to self-improving superintelligence and gambling with our lives," criticizing what he described as a lax approach to safety.

Coxon's resignation and his stated risk assessment highlight growing safety concerns surrounding AI development. Experts in the field have long discussed potential risks associated with AI systems that surpass human intelligence, including loss of control and unintended consequences.

Anthropic has not directly commented on Coxon's specific claims or departure. The company has previously emphasized its commitment to developing AI safely and responsibly. However, the intense competition and rapid pace of advancement in the AI sector raise questions about the maintenance of safety standards.