US Treasury Secretary Threatens Sanctions Against Chinese AI Companies Over Alleged Intellectual Property Theft

Jul 21, 2026

U.S. Treasury Secretary Scott Bessent said the administration would examine Chinese open-source AI models for signs of intellectual property theft and could sanction Chinese AI companies if theft is established. The warning comes as Chinese models such as Moonshot AI's Kimi K3 become increasingly competitive with leading U.S. systems. Anthropic has accused Chinese AI laboratories of illicitly extracting Claude's capabilities through model distillation, although some industry figures dispute that distillation necessarily constitutes theft. The United States and China plan to hold their first official AI dialogue under President Donald Trump in September 2026, although the dates and details have not been finalized.

Treasury sanctions threat

▪Treasury Secretary Scott Bessent said on July 21, 2026, that watermarks from American large language models had been found on many Chinese AI offerings.
▪Axios reported on July 20, 2026, that the Trump administration was considering a wholesale ban on Chinese open-source artificial intelligence models, although others disputed that claim.
▪United States Treasury Secretary Scott Bessent threatened to impose sanctions on Chinese artificial intelligence companies if intellectual property theft is established.

Chinese AI model advances

▪Moonshot AI released Kimi K3, which independent testing cited by Gizmodo found competitive with Anthropic's Fable 5 and OpenAI's GPT-5.6 on several benchmarks.
▪DeepSeek released its R1 model in January 2025, matching common capabilities of OpenAI's top models at the time without requiring advanced Nvidia GPUs.

AI distillation allegations

▪Anthropic accused Moonshot AI, DeepSeek, and MiniMax in February 2026 of violating terms of service to extract Claude's capabilities through distillation.
▪Anthropic accused Alibaba of conducting a distillation campaign against its Claude models by generating over 28.8 million interactions through approximately 25,000 fraudulent accounts between April 22 and June 5, 2026.

September bilateral AI talks

▪The United States and China plan to hold their first official artificial intelligence dialogue under President Donald Trump in September 2026, although the dates and details have not been finalized.
▪Treasury Secretary Scott Bessent is expected to lead the United States delegation during the planned artificial intelligence talks with China in September 2026.

Technology sharing restrictions

▪Gizmodo reported that the Trump administration had limited access to Anthropic's Mythos model to a small group of approved companies and government agencies, making access by Chinese entities unlikely because of national security concerns.
▪The United States government restricted China's access to advanced semiconductor chips and tightened export controls to maintain its lead in artificial intelligence.