US Threatens Sanctions Against Chinese AI Firm Moonshot Over Alleged Theft of AI Models

Accusations of Stealing an American AI Model

Technological tensions between the United States and China in the field of artificial intelligence have intensified after the White House leveled explicit accusations against Chinese company Moonshot, alleging it unlawfully exploited Fable, a model developed by American company Anthropic, through a technique known as "model distillation." US Treasury Secretary Scott Bessent announced that imposing sanctions and adding non-compliant companies to the entity blacklist remains a serious option on the table.

What Is Model Distillation and Why Is It Controversial?

Model distillation is a common technique in AI development that involves training a smaller model using the outputs of a larger, more advanced one. While it is widely used to improve model efficiency, it can become a clear violation of intellectual property rights when carried out covertly and without the permission of the original model's owner.

In this context, Bessent wrote on X: "Open source does not mean open license to steal. When Chinese companies engage in covert distillation operations at an industrial scale that crosses the line into theft, sanctions and entity listing will be on the table."

White House Accusations: More Serious Details

The accusations went beyond the issue of distillation to include potential violations of US technology export controls. Michael Kratsios, the White House's Chief Science and Technology Advisor, stated that Moonshot obtained GB300 servers manufactured by Nvidia — servers belonging to the Blackwell generation that are entirely banned from Chinese access. He indicated that the company may have used these servers in Thailand to train its models, raising legal questions about circumventing US export restrictions.

Doubts About the Official Accusations

Several AI experts, however, have questioned the accuracy or validity of these accusations. They point out that the Fable model was not made publicly available until early July, while Moonshot had already launched its Kimi K3 model — as an open-weights model — just days before. This timeline makes it logically difficult to claim that K3 was fundamentally built on distillation from Fable. K3 attracted significant attention for its advanced capabilities, raising serious questions about the value of the massive investments being poured into the AI race by major American laboratories.

Washington's Debate: Should Chinese Open Models Be Banned?

This crisis has exposed a genuine divide within American political and technical circles. One camp believes it is necessary to impose strict restrictions or an outright ban on the use of Chinese open-source models, citing national security concerns and the need to preserve America's technological edge. Among the most prominent proponents of this view is Dean Ball, former White House AI advisor and current head of strategy at OpenAI.

On the other hand, another camp warns that over-regulation could harm innovation and hinder the growth of the open-source ecosystem within the United States itself.

What Can Be Expected?

All eyes are now on the US government's next steps, with the Treasury Department stating it will examine Chinese open-source models for any violations of American intellectual property rights. Chinese AI companies continue to face mounting pressure — both in terms of potential sanctions and in terms of access to the chips and infrastructure needed to develop their models.

TechCrunch has reached out to Moonshot and the US Treasury Department for official comment and is awaiting their response to these accusations.

✦ بقلم فريق دروب أيديا