Chinese AI Labs Launch Massive Distillation Attacks on Anthropic Claude, Tracking 13M Exchanges

Three Chinese AI labs, DeepSeek, Moonshot, and MiniMax, have been identified conducting industrial-scale data extraction campaigns against Anthropic’s Claude models, generating over 13 million exchanges. These sophisticated distillation attacks aimed to illicitly acquire advanced AI capabilities like coding and reasoning, bypassing safety measures and terms of service.