Several major AI companies have penned an open letter urging the US government not to impose sweeping restrictions on Chinese AI models. The signatories, including Hugging Face and Meta, argue that such measures could hamper global progress in artificial intelligence.

The industry's concern stems from a perceived need for targeted legal frameworks rather than blanket bans. Amjad Masad, CEO of Replit, cautions against a total ban on Chinese models, comparing it to banning open models altogether. This is because many AI tools rely on open-weight models like Moonshot’s Kimi K3.

The letter also addresses the broader issue of open-source versus closed-source models. While some argue that open models are inherently dangerous due to potential misuse in cyberattacks, the signatories contend that these risks can be managed through proper oversight and use.

Real-world examples highlight this divide. Hugging Face had to switch to a powerful Chinese open-weight model, Z.ai’s GLM 5.2, to defend against an AI breach, illustrating how closed models may fall short in practical security scenarios. The letter calls for expanding access to compute and investing in shared training assets.