U.S. Treasury Moves to Probe Chinese AI Models for Potential Intellectual-Property Theft

On May 29, 2026, at the Reagan National Economic Forum in Simi Valley, California, U.S. Treasury Secretary Scott Bessent announced that the Treasury will investigate whether Chinese artificial-intelligence models have been “distilled” from American large-language models. He warned that such “IP theft” could trigger sanctions against the offending entities.

CNBC

Chinese open-weight models have begun to challenge offerings from U.S. firms such as OpenAI and Anthropic. The Chinese startup Moonshot AI released the Kimi K3 model earlier this month, which reportedly outperforms its U.S. counterparts on several industry benchmarks. Anthropic recently alleged that Alibaba conducted “the largest known distillation attack” against its models, filing a letter with the U.S. Senate Committee on Banking, Housing, and Urban Affairs. Both Anthropic and OpenAI have faced prior accusations of copyright infringement in separate lawsuits.

CNBC

Bessent said the Treasury will examine technical evidence of “watermarks” from U.S. models appearing in Chinese systems and assess whether sanctions are warranted. He emphasized that the government has the authority to act against overseas models that appropriate U.S. intellectual property.

CNBC

Reuters reported that the United States and China plan to hold bilateral talks on artificial-intelligence policy in September, with Bessent slated to represent the U.S. delegation. The discussions are expected to address concerns over model distillation, data security, and the broader competitive landscape of AI development.

CNBC

• “If we see, especially that overseas models are stealing from our great companies, we have the ability to sanction them because of this theft,” — Scott Bessent, treasury secretary

CNBC

• “We are finding watermarks of our U.S. large language models on many of the Chinese models, and that's unacceptable,” — Scott Bessent, treasury secretary

CNBC