Already a subscriber? Make sure to log into your account before viewing this content. You can access your account by hitting the “login” button on the top right corner. Still unable to see the content after signing in? Make sure your card on file is up-to-date.

Treasury Secretary Scott Bessent has threatened to sanction Chinese AI companies over intellectual property theft.

Getting into it: Bessent made the threat Tuesday on Fox Business, saying the administration has found watermarks from American models embedded in Chinese ones and framing it as a line it is willing to enforce. “This administration supports open source models, but what we do not support is IP theft,” he 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.”

Bessent added, “We are finding watermarks of our U.S. large language models on many of the Chinese models, and that’s unacceptable.” He also said the administration would be looking at it in the coming days or weeks. He floated one other idea: requiring companies that use Chinese models to disclose that to their customers.

The threat lands as Chinese open-weight models close the gap with the top US labs. Moonshot AI’s Kimi K3, released this month, matches or beats models from OpenAI and Anthropic on some industry benchmarks, and one AI evaluator now has it sitting first for front-end coding, ahead of anything either lab has out.

This all comes as the US and China prepare for their first official AI talks under Trump, expected in September ahead of a Xi Jinping trip to the US on September 24, with Bessent leading the US delegation. Chinese negotiators are reportedly bringing their own short list: whether Washington will clamp down on Chinese open-weight models, what rules it sets for releasing its own frontier systems, and Anthropic’s Mythos, which authorities in Beijing view as a hacking risk.