TLDR
- US officials allege Moonshot AI secretly distilled Anthropic’s Fable model to create Kimi K3.
- Michael Kratsios, White House tech policy chief, publicly made the accusations on social media.
- The Chinese firm allegedly created a hidden infrastructure for large-scale AI model distillation.
- Treasury Secretary Scott Bessent indicated potential sanctions and export controls may be imposed.
- AI experts have raised doubts about whether the accusations are technically feasible given the timeline.
The US government has leveled serious accusations against Moonshot AI, a Chinese artificial intelligence company, claiming it covertly leveraged Anthropic’s proprietary technology to develop its Kimi K3 model, which debuted on July 16.
The allegations came from Michael Kratsios, who heads the White House Office of Science and Technology Policy, in a statement posted to X on Wednesday.
“We have information that Moonshot AI distilled Anthropic’s Fable for the development of its K3 model,” Kratsios stated.
According to Kratsios, the company established a concealed internal infrastructure designed specifically for distilling American AI models on a massive scale. This platform allegedly gave Moonshot AI the capability to rotate between different access points to evade detection.
The White House official further claimed that Moonshot AI significantly increased its computational resources, purportedly purchasing server infrastructure in Thailand for model training purposes.
Since its July debut, Kimi K3 has earned recognition as among China’s most advanced artificial intelligence systems. The allegations reflect mounting anxiety in Washington about Chinese companies leveraging American AI technology to accelerate their domestic development efforts.
The Distillation Debate
The White House’s assertions have not been universally accepted.
Several artificial intelligence experts have raised concerns that the timeframe between Anthropic’s model availability and Kimi K3’s launch is insufficient to account for the claimed distillation process.
Anthropic re-released its Fable 5 system on July 1 following the removal of export controls. Moonshot AI launched Kimi K3 merely 15 days afterward on July 16.
Elie Bakouch, who researches at AI company Prime Intellect, expressed skepticism about the technical plausibility. “I don’t think claiming that K3’s performance comes from Fable distillation makes sense technically,” he noted.
Dean Ball, who leads strategic futures at OpenAI, also shared his perspective. He indicated his belief that distillation alone cannot account for Kimi K3’s capabilities.
The administration has not made public any supporting evidence for its allegations. Moonshot AI has remained silent on the matter as of publication.
Sanctions on the Table
Treasury Secretary Scott Bessent emphasized that the government treats large-scale distillation operations as a significant threat to American interests.
“Open source is not open season on American IP,” Bessent declared.
He indicated that Chinese companies engaging in covert large-scale distillation operations may face economic sanctions and potential inclusion on the Entity List, a restrictive measure that curtails access to American technology.
Kratsios emphasized a clear difference between lawful distillation practices, which produce smaller and more efficient models, and what he characterized as intellectual property theft from US companies.
Anthropic disclosed earlier this month that the US Department of Commerce had removed export limitations on its most advanced systems, including Fable 5 and Mythos 5. The company stated it restored global access immediately thereafter.
The controversy carries added weight as both Moonshot AI and Anthropic have been exploring initial public offerings, introducing a commercial element to the conflict.
The administration’s investigation into the alleged infractions continues.


