Bessent Threatens Sanctions on Chinese AI, Greer Opens Review of Model Propagation贝森特威胁制裁中国AI模型，格里尔宣布审查中国AI传播方式
On Wednesday July 22, Bloomberg reported U.S. Trade Representative Jamieson Greer saying officials are taking a 'very close look at how China is propagating its AI development.' Treasury Secretary Scott Bessent separately stated on Fox Business Tuesday that Washington has the ability to sanction overseas AI models found to have distilled from American IP, adding: 'You can't use counterfeit goods.'
The remarks come directly in the wake of Kimi K3's release, which Swiss security firm Aikido found offers cybersecurity performance 'extremely close' to OpenAI's flagship GPT-5.6 Sol at a fraction of the cost. Chinese open-weight models are gaining commercial traction with major Western enterprises including Airbnb.
The U.S. statements stop short of specific measures and no formal investigation has been announced, but they signal the administration is preparing a regulatory response to the latest wave of Chinese frontier open-weight releases, testing frameworks designed primarily around chip access rather than software distribution.
On Wednesday, Treasury Secretary Scott Bessent stated the US could sanction any foreign AI models determined to have stolen American intellectual property, and suggested companies using Chinese AI could face pressure. US Trade Representative Jamieson Greer said his office is taking 'a very close look at how China is propagating its AI development.' Both statements were delivered separately but landed within the same news cycle.
Bloomberg published a detailed analysis Wednesday morning framing the moment as a new test for Washington's containment approach — arguing that the rapid global competitiveness of Chinese open-source models creates friction that export controls on chips alone cannot resolve. The article notes the challenge of sanctioning software that is freely downloadable.
Neither official named a specific Chinese model, company, or timeline for action. The statements are rhetorical escalation, not yet a regulatory instrument. The next concrete trigger to watch is whether the Commerce Department's Bureau of Industry and Security follows with a rulemaking, or whether USTR opens a formal Section 301 investigation into Chinese AI practices.
On Wednesday, Treasury Secretary Scott Bessent stated the U.S. could sanction foreign AI models found to be stealing American intellectual property, and added that companies cannot use what he called 'counterfeit goods' — a framing that extends liability to end-users of Chinese AI, not merely the developers. The same day, USTR Jamieson Greer said his office is taking 'a very close look' at how China is propagating its AI development globally, signaling a formal trade-probe framing is under active consideration.
The remarks, reported by Bloomberg on July 22, arrive as Chinese open-weight models have gained measurable share on global API platforms. The Bessent language is the most specific threat yet: prior U.S. restrictions targeted hardware exports and investment; sanctions on model IP would create compliance exposure for any enterprise deploying DeepSeek, Qwen, or Kimi in production outside China.
Beijing has not issued a formal response as of this bulletin. The backdrop is Xi Jinping's July 17 WAIC speech urging multilateral AI cooperation and warning against any single nation dominating the technology — a posture the new U.S. language directly contests.
On Wednesday, U.S. Trade Representative Jamieson Greer said Washington is taking 'a very close look at how China is propagating its AI development,' while Treasury Secretary Scott Bessent separately stated that the U.S. could sanction foreign AI models found to be stealing American intellectual property and suggested pressure could fall on companies using Chinese AI.
The two statements — from separate departments on the same day — represent the sharpest public articulation so far of a dual-track approach: trade-law instruments via USTR and financial/sanctions tools via Treasury. Neither official named a specific Chinese lab, model, or company.
The backdrop is a rapid rise in the global competitiveness of Chinese AI models, which Bloomberg notes is 'ringing alarm bells in Washington.' Chinese labs including DeepSeek, Qwen, and Kimi have each posted frontier-level results at a fraction of U.S. pricing, complicating the standard chip-export containment playbook that Washington has relied on since 2022.
On Wednesday July 22, Treasury Secretary Scott Bessent said the U.S. could sanction any foreign AI model determined to have stolen American intellectual property, adding that users of such models would face pressure too. His comments came one day after USTR Jamieson Greer confirmed the administration is conducting a formal review of how China is expanding its AI footprint globally.
The paired statements represent the clearest signal yet that the U.S. is considering applying its existing IP-enforcement and sanctions toolkit to the AI model layer — a step beyond the chip export controls already in place. For Chinese labs like DeepSeek, Qwen, and Kimi that have been aggressively expanding in Southeast Asia, the Middle East, and Latin America, the threat introduces a new category of enterprise-customer hesitancy.
Bloomberg's Wednesday coverage frames the dynamic as a test of the standard containment playbook: whether IP-threat rhetoric and potential sanctions can slow the adoption of Chinese models that are already price-competitive and, in some benchmarks, performance-competitive with U.S. offerings.