Washington Names Open-Weight Chinese AI Models as Sanctions Targets for the First Time

2. Washington Names Open-Weight Chinese AI Models as Sanctions Targets for the First Time

On Tuesday, July 21, Treasury Secretary Scott Bessent said the U.S. government would examine Chinese open-source AI models for signs of intellectual property theft and threatened sanctions against any Chinese AI companies where theft is established. Speaking on Fox Business, Bessent said: "If we see, especially, that overseas models are stealing from our great companies, we have the ability to sanction them because of this theft." The statement, first reported by Bloomberg, follows an Axios report from the day prior that the Trump administration is considering a wholesale ban on Chinese open-source models. Moonshot AI's Kimi K3 was named in the article as a recent example of Chinese models gaining capability and market traction.

The strategic shift here is the target, not the tool. Prior moves in the U.S.-China AI competition focused on chip export controls and hardware access. Naming open-weight model exports as a potential sanctions trigger is a different category of action: it would give Washington a mechanism to restrict enterprise procurement of models like Qwen or Kimi K3 at the software layer, not just the silicon layer. That directly threatens the business model of OpenAI and Anthropic, which have cited Chinese open-source alternatives as a capital-raising risk. It also puts the IP-theft framing under pressure: Hugging Face CEO Clem Delangue argued on TechCrunch's Equity podcast that distillation "is a practice that everyone is doing, including companies in the U.S." and that China's progress reflects strong domestic research teams, not copying.

The legal ground is shakier than the political signal suggests. Model distillation does not have a settled legal definition as theft, and Microsoft CEO Satya Nadella publicly criticized large labs this month for treating distillation restrictions as a default. Meanwhile, Anthropic reached a $1.5 billion settlement after a judge ruled it had illegally downloaded millions of copyrighted books for training. Watch whether Treasury produces a formal IP-theft finding or whether this remains a deterrence signal aimed at slowing enterprise adoption of Chinese open-weight models without a binding legal mechanism to back it up.

Source: US threatens sanctions against Chinese AI models over IP theft