The Trump administration is prepared to impose sanctions on Chinese artificial intelligence firms if it finds evidence that their open-source AI models were developed by copying U.S. laboratories, according to Treasury Secretary Scott Bessent. Bessent stated during a Fox Business appearance that while the administration supports open-source models, it strongly opposes intellectual property 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. U.S. officials have reportedly identified "watermarks" of American AI models within Chinese products, a development Bessent described as "unacceptable."
This warning comes amid rising concerns from U.S. AI leaders, including Anthropic, OpenAI, and Google, about Chinese labs allegedly using a technique called unauthorized distillation. This method involves using an advanced model to train a new one without explicit permission. The issue has gained prominence as powerful open-source AI models from China, such as Moonshot's Kimi K3, become increasingly popular in the U.S., often offered at a significantly lower cost than those from U.S. companies.
Anthropic previously accused Moonshot of employing distillation in February, though direct evidence concerning Kimi K3 has not been confirmed. Bessent indicated that U.S. officials would be actively investigating these claims in the near future.
Bessent, who is slated to lead high-level U.S.-China discussions on artificial intelligence in September, also suggested that U.S. companies utilizing Chinese AI models might be required to provide public disclosures. These remarks follow internal debates within the U.S. government regarding the appropriate response to China's rapid advancements in AI.
The notion that U.S. labs maintain a significant lead over China in AI development has been challenged by the recent successes of Chinese firms like Moonshot and Z.ai, which released an open-source model called GLM-5.2. Reports indicate the Trump administration is exploring various measures to curb Chinese AI development, including potentially adding Chinese AI labs to the Commerce Department's "entity list," which would restrict their operations in the U.S. without a license.
The debate over regulation has also drawn differing perspectives from within the tech industry. OpenAI's head of strategic futures, Dean Ball, suggested that the proliferation of open-source AI could lead to what he termed "full AI communism," a viewpoint that has drawn criticism for potentially stifling fair competition. David Sacks, a White House AI adviser, countered these assertions, characterizing Ball's remarks as an attempt at "regulatory capture" by major AI companies seeking to benefit their own interests.