U.S. Treasury Secretary Scott Bessent said on Tuesday that the Trump administration will investigate whether Chinese AI models have been distilled from American models, threatening sanctions against Chinese AI companies if intellectual property theft is established1,2.

"We've seen a lot of talk about open source models coming and threatening the large language models in the US," Bessent said on Fox Business. "This administration supports open source models, but what we do not support is 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 the U.S. is finding watermarks of U.S. large language models on many Chinese models. Bessent's comments were first reported by Bloomberg.

The threat comes as Chinese open-weight models are gaining ground against leading offerings from OpenAI and Anthropic. Moonshot AI, a Chinese startup, released Kimi K3 earlier this month, which outperforms OpenAI and Anthropic models across some industry benchmarks.

Separately, Axios reported on Monday that the Trump administration is considering a wholesale ban on Chinese open source models, though others have disputed that claim. In April, the White House said it would work closely with AI firms to combat IP theft.

Anthropic sent a letter to the U.S. Senate Committee on Banking, Housing, and Urban Affairs last month alleging that Alibaba Group Holding had carried out the "largest known distillation attack" against it to date. Hugging Face CEO Clem Delangue pushed back on the framing, saying distillation is a practice that everyone is doing, including companies in the U.S., and that it is a "very small factor in the ability to create good models".

The U.S. and China are planning to hold talks around AI in September, with Bessent representing the U.S. during those discussions, according to Reuters.

ANALYSIS The sanctions threat represents a potential escalation beyond existing export controls on chips and equipment, extending enforcement directly to AI model outputs. Sanctions targeting specific Chinese AI companies or models would mark a new category of trade restriction in the AI competition between the two countries.

The backdrop includes broader IP disputes in the AI sector. A federal judge on Monday gave final approval to Anthropic's $1.5 billion settlement of a class action copyright lawsuit brought by authors and book publishers ctx. New York Times sued OpenAI and microsoft in 2023, alleging copyright infringement from using its intellectual property to train models.

ANALYSIS The juxtaposition of Bessent's IP theft allegations against Chinese firms with ongoing domestic copyright disputes involving the same American AI companies highlights the complexity of IP enforcement in the AI industry, where the boundaries of permissible training data use remain contested on multiple fronts.