US Accuses China of 'Stealing' AI Technology as Chinese Models Gain Popularity
China is once again in the spotlight thanks to the sophistication of its AI models. Data from the AI model marketplace OpenRouter shows that many companies from the United States are using low-cost AI models made in China.
This trend appears to vindicate China’s AI industry, which relies on open-source and open-weight approaches. China’s reputation in the field of AI is growing ever brighter, while simultaneously narrowing the gap with frontier AI models from the US, such as those from Anthropic and OpenAI.
After DeepSeek shook the American technology industry, Moonshot has now emerged as the new darling through its latest K3 model. However, the US has once again reacted to the popularity of Chinese AI models.
Moonshot, which is based in China, has been accused of stealing data from Anthropic’s most advanced AI model, Fable. Michael Kratsios, a senior technology official in the Trump administration, said Moonshot had used Fable to help develop K3. Moonshot has also been accused of obtaining advanced AI chips from the US firm Nvidia.
“We have information that Moonshot performed distillation on Anthropic’s Fable for the development of their K3 model,” he said in a post on X, quoted by Reuters on Thursday (23 July 2026).
“The practice of covert, industrial-scale distillation aimed at stealing exclusive US technology and weakening American research is unacceptable,” he added.
US Treasury Secretary Scott Bessent said in a separate post on X on Wednesday (22 July) that he was considering adding Moonshot to a trade blacklist and imposing strict sanctions.
The accusations from the Trump administration are likely to heighten tensions over technology between Beijing and Washington, as companies from both countries race to launch increasingly advanced AI models with broad economic impacts and potential military applications.
Moonshot did not immediately respond to a request for comment.
Liu Chang, a spokesperson for the Chinese Embassy in Washington, called Kratsios’s comments “completely groundless” and added that China respects intellectual property protection.
“The relevant parties in the US should respect the facts, set aside prejudice, and stop disparaging and discrediting China’s achievements in developing its artificial intelligence industry,” he said.
Anthropic’s public policy executive, Sarah Heck, said in a post on X that the theft of US models by Chinese parties “poses a serious national security risk to the United States.”