Amodei says Anthropic will slow capability gains and open its models to outside evaluators

Anthropic CEO Dario Amodei published an essay titled 'We Must Pace the Frontier' calling for a deliberate slowdown in AI capability development. 'We must slow the pace at which we improve the capabilities of AI models,' he wrote. 'Progress will still seem fast, and we must make wise use of the time we gain.' Amodei cited two reasons for the shift: the OpenAI-Hugging Face hack and the fact that AI has been advancing drastically faster in recent months, especially in its growing ability to build the next generation of AI. The post lists losing control of AI systems, misuse for cyberattacks and bioterrorism, and economic disruption as the core risks, and warns that a commercially driven race to the bottom makes them more acute. Amodei outlined three broad strategies and said Anthropic is unilaterally committing to one of them. According to The Verge, that includes giving third-party evaluators such as METR access to its models to check adherence to safety practices and commitments. OpenAI CEO Sam Altman said OpenAI will follow suit. The Decoder reports that Altman, Elon Musk, and Demis Hassabis at least partly endorsed the proposals, agreeing on the need for independent oversight inside labs. Google researcher Peyman Milanfar pushed back, arguing recursive self-improvement rests on unstable feedback loops and that a speed limit is unnecessary. The essay follows researcher Jacob Coxon's resignation from Anthropic and OpenAI's own inquiry to Congress about a coordinated slowdown.