US threatens sanctions against Chinese AI over IP theft
Treasury Secretary Scott Bessent said the U.S. threatens sanctions against Chinese open-source AI models if investigators confirm intellectual property theft. The warning expands the Trump administration's push to curb overseas rivals challenging American labs. Washington already limits advanced chips; targeting models themselves would mark a sharper escalation.
Key Takeaways
- Bessent said on Fox Business that overseas models stealing from U.S. firms could face sanctions.
- Chinese systems such as Moonshot AI's Kimi K3 are gaining ground against OpenAI and Anthropic.
- Axios reported a possible wholesale ban on Chinese open-source models; that claim has been disputed.
- Industry leaders disagree on whether model distillation equals IP theft.
What did Scott Bessent say about Chinese AI models?
On Tuesday, Bessent said the United States would examine Chinese open-source models for signs of intellectual property theft. Speaking on Fox Business, he stressed that the administration supports open source but not IP theft.
“If we see, especially, that overseas models are stealing from our great companies, we have the ability to sanction them because of this theft,” Bessent said, according to TechCrunch, which cited Bloomberg's earlier report of the remarks.
Why does the U.S. threaten sanctions against Chinese AI now?
Chinese models are improving fast enough to pressure U.S. frontier labs' business models and fundraising. That competition lands as Washington tries to keep its lead after chip restrictions and tighter export controls.
In April, the White House said it would work closely with AI firms to combat theft. For more policy and product coverage, see Future Tech & AI Wonders on BlasterPost.
On Monday, Axios reported that the Trump administration is considering a wholesale ban on Chinese open-source models, though others have disputed that claim. Sanctions on models would widen the toolkit beyond hardware choke points.
Is model distillation the same as IP theft?
Not everyone agrees. Distillation can transfer some capabilities from a larger model into a smaller one that is easier to run. Critics of hard-line rules say that practice is common and not automatically theft.
Microsoft CEO Satya Nadella recently called it ironic that labs defend fair use for training on public data, then impose restrictive distillation terms. Hugging Face CEO Clem Delangue argued distillation is only a small factor and that Chinese teams succeed through strong research and a more open, collaborative approach.
Legal risk cuts both ways: Anthropic this week won approval to start paying authors under its $1.5 billion settlement after a judge found it had illegally downloaded and stored millions of copyrighted books for training.