U.S. Considers Sanctions Over Alleged Chinese AI Model Theft

Treasury Secretary Scott Bessent announced that the White House is investigating whether Chinese artificial intelligence models were distilled from American technology. Bessent stated that watermarks of U.S. large language models have been found on Chinese versions, which he called unacceptable. He warned that the administration would impose sanctions if overseas models are found to be stealing from U.S. companies.

These tensions follow the release of Kimi K3 by Moonshot AI and the rise of DeepSeek, both of which have rivaled U.S. offerings from OpenAI, Anthropic, and Google. The advancement of Chinese AI contributed to a semiconductor stock sell-off. In response, China is planning a $295 billion investment in nationwide data center infrastructure while alleging that U.S. semiconductors contain backdoor features designed to compromise Chinese models.

Internal U.S. debate has intensified over the government's approach to the AI race. A June 2026 executive order from President Donald Trump required reviews of advanced AI models before release, leading OpenAI and Anthropic to restrict the rollout of certain flagship models. Investor David Sacks criticized these regulatory hurdles and data center restrictions, arguing that they undermine the permissionless innovation that previously allowed the U.S. to lead in technology.