In this press release, China’s Ministry of Commerce flatly rejects the notion that Chinese AI companies have distilled from U.S. AI models, while simultaneously arguing that model distillation is a widespread practice in both countries. This statement is the first direct Chinese government response to U.S. accusations that Chinese AI company Moonshot AI distilled from leading U.S. AI models to build its latest model, Kimi K3. Moonshot released Kimi K3 on July 16, 2026, claiming it achieves world-class performance on several benchmarks.
An archived version of the Chinese source text is available online at: https://perma.cc/VPT7-K229
Ministry of Commerce Press Spokesperson Answers Reporter’s Question about the United States Threatening to Launch Investigations and Implement Sanctions Against Chinese Artificial Intelligence Enterprises
Question: Recently, some high-level U.S. government officials stated that they will launch an investigation into Chinese artificial intelligence (AI) enterprises’ so-called “distillation” of U.S. frontier models, and, using “stealing American intellectual property (IP),” etc., as a justification, they might sanction Chinese enterprises. What does the Chinese side have to say about this?
Answer: We are paying attention to this matter. The Chinese side has consistently opposed the politicization and instrumentalization (工具化) of technology trade issues, and the use of all manner of baseless excuses to smear Chinese enterprises and implement sanctions against them. The U.S. side has ignored such facts as the extremely short interval between the release of Chinese enterprises’ AI models and U.S. frontier models, and the fact that some Chinese models already have world-leading capabilities, and has accused them of relying on so-called “distillation” and “stealing American IP,” and has also threatened to implement sanctions to suppress them. These actions lack any actual evidence, have no legal backing, and effectively amount to a double standard. It is a classic case of hegemonic behavior in the AI sphere.
Innovation is not the exclusive preserve of any one side. Chinese AI enterprises have diligently undertaken basic research over many years, and have persisted in giving equal weight to technological self-reliance (自立自强) and open cooperation. They have upheld the principles of joint consultation, joint construction, and mutual benefit, they have advanced the rapid development of AI technology, and they have achieved gratifying successes in more and more fields such as front-end coding capabilities. As we understand it, many U.S. AI companies have distilled from China’s models for their own R&D and training. Moreover, many enterprises in U.S. industry circles have openly opposed U.S. threats of sanctions against Chinese AI enterprises. Nearly 200 U.S. startups urged the U.S. government not to cut off their access to Chinese open-source models, believing that this would weaken the competitiveness of U.S. enterprises. Some large U.S. multinationals have said that model distillation is a technology that is widely used in the industry, and that U.S. enterprises should be permitted to use Chinese open-source AI models.
The Chinese side urges the U.S. side to listen attentively to objective, rational voices in the industry in both China and the United States, rectify its hegemonic thinking, and stop its defamation and threats of sanctions against Chinese enterprises. The Chinese side will take all measures necessary to staunchly defend its legitimate and legal rights and interests against any action that substantially harms the interests of the Chinese side. We hope that the U.S. side earnestly puts into practice the consensus reached at the meeting of the two countries’ heads of state, moves forward together (相向而行) with the Chinese side, strengthens dialogue and communication on AI, works together with us to promote AI for the good of society (人工智能向上向善), and benefits both countries and the entire world by promoting and improving AI governance.
To download a pdf version of this translation, click the link below.