Bessent Says US Considering Sanctions on China Over AI Open-Weight Model Theft

Treasury Secretary Scott Bessent announced the U.S. is considering sanctions on China over alleged theft of American large language models through model distillation. Bessent specifically named Chinese open-weight AI models like Moonshot's Kimi, DeepSeek, GLM of z.ai, and Alibaba as targets. In a Fox Business interview, Bessent framed the potential sanctions as an anti-piracy measure, stating 'You can't use counterfeit goods.' Possible sanctions include Specially Designated Nationals (SDN) financial sanctions that would freeze U.S.-based assets and bank accounts, or orders for U.S. companies and cloud providers to audit and remove Chinese models containing stolen weights. The announcement comes ahead of a late-September AI Summit between China and the U.S., which Bessent is leading. The article notes that nearly 50% of weekly API tokens for U.S. users on OpenRouter came from Chinese open-weight models, and suggests U.S. alternatives like Google's Gemini 3.6 Flash, Meta's Muse Spark 1.1, and xAI's Grok 4.5 could benefit if sanctions are imposed.