US Accuses China's Moonshot AI of Stealing Technology from Anthropic

Michael Kratsios, a White House science advisor, accused China's Moonshot AI of a large-scale effort to steal capabilities from Anthropic's Fable LLM to develop its Kimi K3 model. The allegations suggest that Moonshot used distillation techniques, querying Fable to replicate its capabilities. Additionally, Kratsios claimed that Moonshot accessed restricted Nvidia servers to train K3, raising concerns about U.S. technology being compromised. The Kimi K3 model, unveiled recently, has been noted for its advanced capabilities, potentially narrowing the gap between Western and Chinese AI models. The U.S. government is considering sanctions against Chinese AI models if evidence of theft is confirmed. Moonshot plans to release Kimi K3 as an open-source model on July 27, 2026. The situation has prompted increased scrutiny of Chinese AI companies and their practices. Anthropic previously accused Alibaba of similar technology theft, urging Congress to take action.

Key Points: • Moonshot AI is accused of stealing capabilities from Anthropic's Fable LLM. • The Kimi K3 model is set to be released as an open-source model on July 27, 2026. • U.S. officials are considering sanctions against Chinese AI models if theft is confirmed.