Anthropic’s Dario Amodei responds: doesn’t oppose open-weight models, but fears Chinese AI

Anthropic CEO Dario Amodei has publicly addressed industry speculation regarding his company's stance on open-weight AI models, specifically refuting claims that Anthropic supports bans on such technology. Amodei emphasized that open-weight models, unless possessing dangerous capabilities, serve as a public good, providing value to businesses, developers, and researchers without significant cost beyond computational resources.

His clarification follows an open letter from Nvidia and other AI companies urging policymakers against premature restrictions on open-weight AI. This industry debate has frequently centered on allegations of Chinese AI labs advancing their capabilities through intellectual property theft, often employing methods like 'distillation' to learn from existing models.

While not opposing open-weight models in general, Amodei articulated long-standing fears regarding authoritarian governments, with China being the most capable, developing advanced AI. He expressed concern that such AI could lead to permanent military superiority or be used to repress populations. He also fears AI enabling biological attacks, where open-weight models pose risks due to difficulty in applying guardrails or monitoring usage.

To counter these threats, Amodei advocates for restricting China's access to powerful chips and cracking down on distillation. He also supports the establishment of a global model safety testing organization, potentially involving China, believing that limited cooperation on preventing AI biological weapons could be mutually beneficial.