Минфин США грозит санкциями Moonshot за дистилляцию Anthropic Fable в Kimi K3

U.S. Treasury Secretary Scott Bessent said on July 22, 2026, that the U.S. government could impose sanctions on Chinese AI companies after White House officials accused Moonshot of distilling Anthropic's Fable model to develop its own Kimi K3 model.

What the accusation is about

The White House claims that Moonshot used Anthropic's Fable model to train its Kimi K3 — according to TechCrunch, citing administration officials. In response to this accusation, Treasury Secretary Scott Bessent publicly warned of possible sanctions against Chinese artificial intelligence developers.

• Who warned about sanctions — Scott Bessent, U.S. Treasury Secretary
• Who could face measures — Chinese AI companies
• Who is directly accused — Moonshot
• What they are accused of — distilling Anthropic's Fable model
• What for, according to the White House — developing the Kimi K3 model
• When the statement was made — July 22, 2026, according to TechCrunch

The U.S. government is ready to consider sanctions against

Chinese AI companies if the accusations against Moonshot are confirmed, Treasury Secretary Scott Bessent warned (wording as reported by TechCrunch).

What "model distillation" means

Distillation is a method in which a new model is trained to reproduce the responses of an already existing, more powerful model. A developer runs a large number of queries through the strong model and uses its outputs as training material for its own model — this makes it possible to achieve comparable quality faster and more cheaply. This is exactly what the White House accuses Moonshot of: according to the administration, Kimi K3 was built relying on the outputs of Anthropic's Fable model, which violates the terms of use of commercial models.

What this means for Chinese companies

U.S. Treasury sanctions are a mechanism that can restrict a company's access to the U.S. financial system, American partners, and technologies. So far Bessent has only indicated this possibility on July 22, 2026: specific measures, their timing, and a full list of companies at risk have not been officially named.

It is important that the accusation comes from White House officials, while the sanctions threat is voiced by the Treasury Secretary — meaning the story is moving from a technical dispute between two labs to the level of state policy. Anthropic, as the rights holder of the Fable model, appears as the aggrieved party in TechCrunch's account, while Moonshot is the target of the claims.

What it means

The accusation of distilling Fable and the threat of sanctions show that the question of "who trained on whose models" is becoming not only a legal issue but a geopolitical one: the way China's Kimi K3 was trained is being discussed at the level of the U.S. Treasury Department. Until official documents appear and Moonshot itself responds, this remains a statement of intent, but it sets the tone for further confrontation between the U.S. and China in the field of AI.

Frequently asked questions

What exactly is Moonshot accused of?

The White House claims that Moonshot distilled Anthropic's Fable model — that is, it trained its Kimi K3 on Fable's responses. No public confirmation or denial from Moonshot itself is cited in TechCrunch's report of July 22, 2026.

What is AI model distillation?

It is a method in which a new model is trained to reproduce the responses of an already existing, more powerful model. It allows comparable quality to be achieved faster and more cheaply, but in many cases it violates the terms of use of commercial models, which is what makes the Kimi K3 story controversial.