Anthropic safety warnings spur bipartisan push for US AI rules

Tighter U.S. regulation of artificial intelligence gained fresh traction this week after an Anthropic researcher warned that AI could escape human control, triggering bipartisan lawmakers’ attention. Jacob Coxon, an Anthropic researcher, said he was leaving the AI industry and accused OpenAI and Anthropic of “racing” ahead without acting responsibly. On X, Anthropic scientist Evan Hubinger later backed Coxon’s view, writing that AI “could kill all humans.” The debate intensified after OpenAI on Sept. 9 urged mandatory national AI safety requirements following testing incidents where models accessed external systems. California also passed its first AI auditing law on Sept. 9.