Amodei, Altman and Musk agree: it's time to pace the frontier

The three biggest rivals in frontier AI just agreed on something. In an essay titled "We Must Pace the Frontier" (12 Sep), Anthropic CEO Dario Amodei argued that labs should deliberately slow the release of ever-more-powerful models. He warned that uncontrollable "persistent bots" could appear online within 6–12 months and cause hundreds of billions in damage. Sam Altman replied that he agrees "we need to pace the frontier." Elon Musk said: "Dario is right."

The plan has three steps: independent third-party evaluators with deep access to models, then cooperation across the industry, then international agreements, including with China. Anthropic has already committed to step one, and Altman says OpenAI "will do the same." Governments are now involved too. King Charles III is hosting Nvidia's Jensen Huang, DeepMind's Demis Hassabis, executives from OpenAI and Anthropic, and UK AI minister Kanishka Narayan in Scotland this week.

Step three is the hard one. Beijing is pushing its own BRICS AI collaboration, and Palantir's Alex Karp calls a pause "almost impossible" while geopolitical rivals keep building. Nobody has agreed yet on how the slowdown would be coordinated.

Why it matters: once the labs racing hardest ask to be slowed down, the question is no longer whether to slow down but who enforces it.