Anthropic declines to sign while Meta speaks on two tracks: unity and division in U.S. AI over open weights

Anthropic CEO Dario Amodei said the company does not want to ban open-weight models, but supports tighter controls on frontier models based on capability and mandatory safety testing.

Meta CEO Mark Zuckerberg argued that AI should not be concentrated in a few hands, opposed banning Chinese AI models, and backed an open-weight ecosystem and checks on power.

The contrasting positions highlight a split in the U.S. AI industry over whether to use country-of-origin bans or capability-based rules for Chinese open-weight models.

The debate could shape upcoming BIS rules, affect the reach of Chinese models such as Qwen and DeepSeek, and influence the competitive strategies of Anthropic, Meta, and others.