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

What the report says

TechCrunch reported that Anthropic founder and CEO Dario Amodei published a response Monday addressing industry concerns that Anthropic supports a U.S. government ban on open-weight AI models, including models from China. Amodei wrote that the company has not pushed for such a ban and said open-weight systems without dangerous capabilities can benefit businesses, developers and researchers.

The response followed an open letter shared Friday by Nvidia CEO Jensen Huang and signed by companies including Hugging Face, Meta, Microsoft, Mistral and Nvidia. That letter urged policymakers to avoid broad early restrictions on open-weight AI. TechCrunch said the wider debate has become tied to concerns that Chinese AI labs are advancing partly through alleged intellectual property theft from U.S. companies, including through “distillation,” a method in which one model is queried to infer how it works.

Amodei drew a distinction between ordinary use of open-weight models and what he sees as risks from authoritarian governments developing highly capable AI. According to TechCrunch, he expressed concern that such systems could aid military dominance, domestic repression, cyberattacks or biological attacks. He argued open-weight models can be harder to control once released because safeguards and monitoring are more difficult to apply.

TechCrunch reported that Amodei supports measures such as limiting China’s access to advanced chips, cracking down on distillation and building a global safety testing regime for the most powerful models. He said such testing would need participation from all major countries, including China, and suggested limited cooperation on preventing AI-enabled biological weapons may be possible because it would serve shared interests.