On Tuesday, Treasury Secretary Scott Bessent stated the U.S. would examine open source models from China for signs of intellectual property theft, threatening sanctions against Chinese AI companies if IP theft is established.
“We’ve seen a lot of talk about open source models coming and threatening the large language models within the U.S.,” Bessent stated on Fox Business Tuesday. “This administration helps open source models, but what we do no longer help is IP theft. If we see, mainly, that overseas models are stealing from our great companies, we’ve got the ability to sanction them because of this theft.”
Bessent’s comments were first reported by Bloomberg.
The assertion comes as Chinese models — most lately Moonshot AI’s Kimi K3 — are gaining in capabilities and popularity, threatening to harm the business models of top American AI firms like OpenAI and Anthropic, as well as their abilities to elevate more capital to persist creating frontier models.
On Monday, Axios stated that the Trump administration is evaluating a wholesale ban on Chinese open source models, even though others have disputed that claim.
AI companies had been warning for months against campaigns by foreign actors to copy their AI technology and redeploy it as open source. In April, the White House stated it would work precisely with AI corporations to combat the theft.
Sanctions from the U.S. against Chinese models would add to the developing listing of strategies the authorities is trying to hold the lead in the AI race. After limiting China’s access to advanced chips and tightening export controls, Washington is now signaling it may target the AI models themselves, a move that might mark a considerable escalation in the technological competition between frontier labs and Chinese open source options.
Model distillation is a method that permits some of a larger model’s capabilities to be translated into a smaller systems that’s simpler to run — but not everyone agrees that distilling another company’s model constitutes theft.
Earlier this month, Microsoft CEO Satya Nadella denounced large labs for making just this assumption: “While the great innovation that comes from model companies having fair use rights to train models on public data is required, I find it ironic that the status quo is to then turn around and impose restrictive terms on distillation.”
AI labs’ training practices persist to be a source of legal risk for the companies. Anthropic this week got the green light to begin cutting authors checks as part of its $1.5 billion settlement after a judge ruled it had illegally downloaded and stored millions of copyrighted books to train its AI.
Moreover, some in the industry claim that distillation isn’t the only purpose China is catching up to U.S. AI companies.
“We know distillation to be a very small component in the ability to create suitable models, and it’s a exercise that everyone is doing, including companies within the U.S.,” Hugging Face CEO Clem Delangue stated on a latest episode of TechCrunch’s Equity podcast. “If it were easy just to do distillation to get good at building AI models, there would be many other nations, inclusive in the U.S., with a much better open source AI. The reality is they have genuinely, actually good research teams in China…taking a much more open and collaborative approach to AI than in the U.S.”
