Anthropic CEO calls for slowing the pace of frontier AI capability growth
Anthropic CEO Dario Amodei has urged frontier AI companies to slow the rate at which model capabilities advance, arguing that the industry needs more time for safeguards, alignment work, and independent evaluation.
Amodei outlined a three-step framework for pacing frontier AI development rather than stopping it. He called for permanent third-party reviewers with access to relevant internal risk-assessment processes, voluntary industry standards to reduce race-to-the-bottom incentives, and broader coordination on AI safety. He said technical progress should continue, but at a pace that gives companies and evaluators more time to manage risks.
A public call from the head of a leading frontier AI lab to deliberately pace capability growth raises the pressure for stronger evaluation and oversight across the industry. If similar ideas are adopted by other labs or regulators, they could affect model release schedules, external testing requirements, and governance for the most capable systems.
Maybe — This is not a product release, but teams building on frontier models should watch for stricter evaluation, deployment, and governance requirements.