US Accuses Moonshot AI of Distilling Anthropic’s Fable to Build Kimi K3

The White House has accused Chinese AI startup Moonshot AI of using Anthropic’s Claude Fable model to develop its newly launched Kimi K3, alleging the company carried out large-scale model distillation and used infrastructure designed to avoid detection. 

In a post on X, White House Office of Science and Technology Policy Director Michael Kratsios said the US had information indicating that Moonshot AI “distilled Anthropic’s Fable for the development of its K3 model.”

“To do this they developed a sophisticated internal platform to conduct large scale distillation against US models, allowing them to quickly switch between multiple methods of access to avoid detection,” Kratsios wrote.

The accusation comes days after Moonshot unveiled Kimi K3, an open-weight model that drew attention for approaching the performance of leading US frontier models.

He also alleged that Moonshot AI had acquired servers equipped with NVIDIA GB300 GPUs and had accessed GB300 systems in Thailand “likely to train its AI models.”

Kratsios distinguished between legitimate model distillation and what he described as industrial-scale theft of proprietary technology.

“The United States strongly supports the free and fair development of AI,” he said. “Legitimate AI distillation used to create smaller, more efficient models plays a vital role in this open innovation ecosystem. However, large-scale, covert industrial distillation aimed at stealing proprietary US technology and undermining American research is unacceptable.”

The US government has not released technical evidence supporting the allegations, and Moonshot AI had not publicly responded at the time of publication.

The accusations come less than a week after Moonshot AI launched Kimi K3, a 2.8 trillion-parameter open-weight mixture-of-experts model with a one million-token context window. 

The company said K3 was built for reasoning, coding and agentic workloads, and positioned it against Anthropic’s Fable 5 and OpenAI’s GPT-5.6 series. 

The allegations also revive concerns over AI model distillation, a technique in which a smaller model learns from the outputs of a larger one. The approach is widely used across the industry for efficiency and cost reduction. 

However, AI companies have increasingly argued that using proprietary frontier models without permission to create competing systems crosses legal and commercial boundaries.

Anthropic’s Fable 5 has itself been at the centre of recent US policy actions. The model was temporarily restricted last month after the US government imposed export controls citing national security concerns before restoring global access on July 1 after those controls were lifted.