U.S. Treasury Secretary Scott Bessent told Fox Business on July 21 that the administration will examine whether Chinese AI systems have been created by copying from U.S. firms through a process known as distillation. Bessent said, "Investigators will look into the issue in the coming days or weeks." If they see that Chinese models were built by copying from U.S. companies, the government has the power to sanction them.
"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.
Bessent explained that the process of copying is known in the industry as distillation. He emphasized that such duplication undermines the competitive edge of American innovators. These watermarks, which are subtle signature patterns embedded in model outputs, can serve as evidence of unauthorized copying. He said the administration is finding watermarks of U.S. large language models on Chinese models.
Get the market news that matters in a five-minute read with Market Briefs, our free daily newsletter
"We are finding watermarks of our U.S. large language models on many of the Chinese models, and that's unacceptable," he said.
In June, Anthropic alleged to the Senate Banking Committee that the Chinese company Alibaba had carried out what it described as the largest distillation attack ever mounted against its models.
Earlier this month, Chinese startup Moonshot AI released a model called Kimi K3 that outperforms American companies OpenAI and Anthropic across some industry benchmarks. The emergence of high-performing Chinese models has spurred debate over whether U.S. export controls are sufficient to protect American AI intellectual property.
Anthropic and OpenAI themselves have also been accused of intellectual property violations. In 2023, The New York Times filed a lawsuit against OpenAI and Microsoft, claiming they unlawfully used its copyrighted articles to train their AI systems. Last September, Anthropic reached a $1.5 billion settlement in a class action brought by authors who accused the firm of downloading books from unauthorized sources.
The allegations arrive amid a broader geopolitical contest to dominate artificial intelligence. The stakes are enormous, as AI leadership could confer economic and military advantages for decades. The planned September 2026 talks between U.S. and Chinese officials represent a diplomatic effort to address these disputes, though Bessent faces pressure from lawmakers who argue that existing export controls - particularly on advanced semiconductors - may not be enough to stop intellectual property leakage via distillation, which requires less computing power than training models from scratch.
China has rapidly advanced its own large language models - such as DeepSeek and Kimi - prompting U.S. officials to weigh export controls, tariffs, and now potential sanctions. The planned September talks represent a diplomatic effort to address these disputes, though tensions remain high as both sides accuse each other of unfair practices. Analysts say the outcome could shape global AI supply chains and determine which country sets the standards for future development.
Join Market Briefs, our free daily newsletter, for a quick daily rundown of the markets
