Anthropic’s CEO wins backing for slowing frontier AI

Dario Amodei, the CEO of Anthropic, is asking AI labs to slow the rate at which they make their most powerful models more capable—not to stop building them. His argument is that these systems are beginning to help create the next generation of AI, so capability gains could outrun the testing, safeguards, and human oversight meant to keep them controllable. He proposed giving independent safety teams continuing access inside leading labs, then building common limits and international coordination; Anthropic says it will adopt the first measure, and OpenAI CEO Sam Altman said OpenAI would too. Elon Musk of xAI also backed the call, an unusual public alignment among rivals, though it is still a voluntary direction rather than a binding industry deal.