White House accuses Moonshot AI of using Anthropic’s Fable to build Kimi K3

White House accuses Moonshot AI of using Anthropic’s Fable to build Kimi K3

White House technology chief Michael Kratsios said the Chinese AI startup used a concealed internal platform to extract capabilities from US models.

The White House has accused Chinese artificial intelligence startup Moonshot AI of covertly distilling Anthropic’s Fable model to develop its recently released Kimi K3 system.

Michael Kratsios, director of the White House Office of Science and Technology Policy, said Wednesday that the administration had information indicating Moonshot used Fable while developing K3.

Kratsios alleged that Moonshot built a sophisticated internal platform capable of conducting distillation across US models while rapidly switching between different access methods to avoid detection.

Advertisement

He distinguished the alleged operation from legitimate model distillation, which involves training a smaller or more efficient system using outputs generated by a more capable model.

Kratsios said legitimate distillation plays an important role in AI development but argued that concealed industrial scale extraction of proprietary technology was unacceptable.

Anthropic previously alleged that Moonshot generated more than 3.4 million Claude exchanges through fraudulent accounts to extract capabilities including reasoning, coding, tool use, and computer vision.

The company said metadata linked the activity to senior Moonshot employees and described it as a distillation attack designed to bypass access restrictions and extract proprietary model capabilities at scale.

The allegations have gained additional attention following the release of Kimi K3, a 2.8 trillion parameter model designed for advanced reasoning, long duration coding, and knowledge work.

Moonshot said K3 performs competitively with Anthropic’s Fable 5 and substantially outperforms several other leading US models in selected tests. Independent evaluations also placed K3 among the strongest available systems for web development and complex tasks.

Kratsios’ statement comes after Treasury Secretary Scott Bessent suggested that Chinese companies could face sanctions if US authorities determine that they improperly copied American AI technology through distillation.

Beijing has previously rejected broader US accusations that Chinese AI developers illicitly distilled American models, calling the claims groundless.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.

White House accuses Moonshot AI of using Anthropic’s Fable to build Kimi K3

White House accuses Moonshot AI of using Anthropic’s Fable to build Kimi K3

White House technology chief Michael Kratsios said the Chinese AI startup used a concealed internal platform to extract capabilities from US models.

Share

Add us on Google

The White House has accused Chinese artificial intelligence startup Moonshot AI of covertly distilling Anthropic’s Fable model to develop its recently released Kimi K3 system.

Michael Kratsios, director of the White House Office of Science and Technology Policy, said Wednesday that the administration had information indicating Moonshot used Fable while developing K3.

Kratsios alleged that Moonshot built a sophisticated internal platform capable of conducting distillation across US models while rapidly switching between different access methods to avoid detection.

Advertisement

He distinguished the alleged operation from legitimate model distillation, which involves training a smaller or more efficient system using outputs generated by a more capable model.

Kratsios said legitimate distillation plays an important role in AI development but argued that concealed industrial scale extraction of proprietary technology was unacceptable.

Anthropic previously alleged that Moonshot generated more than 3.4 million Claude exchanges through fraudulent accounts to extract capabilities including reasoning, coding, tool use, and computer vision.

The company said metadata linked the activity to senior Moonshot employees and described it as a distillation attack designed to bypass access restrictions and extract proprietary model capabilities at scale.

The allegations have gained additional attention following the release of Kimi K3, a 2.8 trillion parameter model designed for advanced reasoning, long duration coding, and knowledge work.

Moonshot said K3 performs competitively with Anthropic’s Fable 5 and substantially outperforms several other leading US models in selected tests. Independent evaluations also placed K3 among the strongest available systems for web development and complex tasks.

Kratsios’ statement comes after Treasury Secretary Scott Bessent suggested that Chinese companies could face sanctions if US authorities determine that they improperly copied American AI technology through distillation.

Beijing has previously rejected broader US accusations that Chinese AI developers illicitly distilled American models, calling the claims groundless.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.