Anthropic CEO Dario Amodei has set the record straight: his company never pushed for a ban on open-weight AI models. In a Monday blog post, he pushed back against industry chatter that Anthropic supported U.S. government restrictions on Chinese open-weight models, or open-weight models in general. “Anyone who has read my past writing should know that I don’t regard such bans as a useful measure,” he wrote. His statement came after Nvidia’s Jensen Huang and a coalition of AI firms—including Meta, Hugging Face, Microsoft, and Mistral—urged policymakers to avoid premature restrictions on open-weight AI.

The real threat, Amodei argues, isn’t open-weight models themselves but how Chinese labs use them—specifically through model distillation, where one AI bombards another with queries to reverse-engineer its behavior. He sees open-weight models without dangerous capabilities as a public good, providing value to developers and researchers. But he warns that authoritarian governments could build more powerful models for military superiority or repression. Amodei supports existing U.S. chip restrictions on China and wants a formal crackdown on distillation as a form of intellectual property theft.

Amodei also backs global safety testing for the most capable models, regardless of origin or whether they’re open or closed. He notes that even the Chinese government might cooperate if testing prevents biological weapon risks—a shared interest. This runs counter to open-source advocates who argue that broad access helps defenders. For data and ML engineers, the debate underscores a growing tension between open model benefits and the security risks of distillation, with policy responses likely to shape how models are shared and deployed.

Source: TechCrunch