U.S. Officials Accuse Moonshot AI of Distilling Anthropic Model for Kimi K3

AI developers often use model distillation to train smaller or more efficient systems from the outputs of larger models. The practice becomes contentious when conducted without authorization, raising questions over terms of service, intellectual property and controls on access to advanced AI. Allegations involving China’s Moonshot AI and U.S.-based Anthropic therefore carry significance beyond a single product, feeding a wider debate over technology transfer and intensifying competition between the United States and China.

U.S. officials have accused Moonshot AI of extracting outputs from Anthropic’s Fable model without permission and using them to distill Kimi K3, according to the report. The available account does not specify the volume of outputs taken, a monetary value or the exact dates of the alleged activity. The case has prompted discussion among major technology companies, including OpenAI and Nvidia, over model safeguards, competitive exposure and how regulators should respond to unauthorized distillation.