Anthropic says they’re OK with open models. The real danger is China.

In a blog post published on July 27, Anthropic CEO Dario Amodei addressed an open letter signed by Nvidia, Meta, Microsoft, and Mistral that criticized efforts to regulate open-source AI models. According to Amodei, his company does not advocate for banning open-weight models. When these models lack dangerous capabilities, they are essentially free goods.

1The Real Fear Is China, Not Open Weights

Amodei believes the real danger comes from authoritarian regimes, particularly the CCP. He is concerned that China will be able to build powerful enough models to secure “permanent military dominance” over other countries and suppress its own citizens, potentially through biological attacks. Instead of banning open weights, Amodei thinks it’s more important to limit the export of high-performance chips, ban model distillation from cutting-edge systems, and create an international organization to review AI safety that involves China.

This announcement is significant because it represents a shift in the conversation surrounding AI regulation. Rather than discussing whether AI should be protected by copyright law, the debate has moved on to who should have access to the world’s most advanced AI models and under what circumstances.