US Threatens Sanctions on Chinese AI Models as Kimi K3 Intensifies IP Theft Fears
Updated
Updated · TechCrunch · Jul 21
US Threatens Sanctions on Chinese AI Models as Kimi K3 Intensifies IP Theft Fears
3 articles · Updated · TechCrunch · Jul 21
Summary
Scott Bessent said the Treasury will examine Chinese open-source AI models for intellectual property theft and could sanction companies if theft is established.
The warning targets a fast-rising class of Chinese models, including Moonshot AI’s Kimi K3, that U.S. officials and labs say could undercut OpenAI and Anthropic and weaken their ability to raise capital.
Washington has already tightened chip and export controls on China; targeting the models themselves would mark a sharper escalation in the U.S.-China AI contest.
Distillation remains contested as a legal and technical issue: Microsoft’s Satya Nadella questioned treating it as theft, while Anthropic is beginning payments under a $1.5 billion copyright settlement over its own training practices.
Some industry figures, including Hugging Face CEO Clem Delangue, argue China’s gains reflect strong research teams and a more open development approach, not just copying U.S. systems.
Are U.S. firms fueling China's AI rise by adopting the very models the government claims are stolen?
Can U.S. sanctions on AI software succeed where hardware export controls have seemingly fallen short against China?
Is China's AI 'distillation' a new form of industrial espionage or just aggressive technological competition?
2026 U.S.-China AI Conflict: Sanctions, Model Distillation, and the Future of Global Tech Governance
Overview
In July 2026, the U.S. government ramped up scrutiny of alleged intellectual property theft in the AI sector, driven by concerns over America’s ability to maintain its lead in the global AI race. This comes as Chinese open-weight models, like those from DeepSeek, rapidly advance and challenge top U.S. offerings. The debate has moved beyond traditional IP disputes, now raising national security issues and fears of export control circumvention. OpenAI has accused Chinese startups of using new, hidden distillation methods to extract data from ChatGPT, highlighting the escalating competition and the urgent need for stronger safeguards.