Anthropic CEO Dario Amodei has clarified that his company has never advocated for banning open-weight AI models, responding to an open letter from Nvidia, Hugging Face, Meta, Microsoft, and Mistral urging policymakers not to impose broad "premature restrictions" on open-weight models. Amodei stated that open-weight models without dangerous capabilities are a "public good" that provide value to businesses, developers, and researchers.
However, Amodei expressed longstanding fears about "authoritarian governments" building AI more powerful than U.S. models to achieve "permanent military superiority" or use AI to repress their own people. He identified the Chinese Communist Party (CCP) as the "most capable" authoritarian government in this regard, while noting he fears AI will enable "biological attacks" in addition to cybersecurity threats.
Amodei argued open-weight models pose greater danger in biological attack scenarios because guardrails and usage monitoring are difficult to apply once released. He recommended restricting China's access to powerful chips and called for a crackdown on distillation—the practice where AI models are bombarded with prompts to learn how another model works. He also voiced support for growing U.S.-led efforts to address these concerns.