Anthropic accuses DeepSeek, other Chinese AI developers of 'industrial-scale' copying — Claims 'distillation' included 24,000 fraudulent accounts and 16 million exchanges to train smaller models

Anthropic has accused Chinese AI developers DeepSeek, Moonshot, and MiniMax of engaging in large-scale distillation, using 24,000 fraudulent accounts and 16 million exchanges to train smaller models with Anthropic's Claude capabilities. While distillation is a legitimate machine learning technique, when used illicitly, it can violate U.S. export controls and end-user agreements. The Chinese developers allegedly used commercial services to access frontier models and created networks of fraudulent accounts to extract capabilities. Anthropic is enhancing defenses to detect and prevent such distillation attempts and is collaborating with other AI labs and authorities to address the issue on an industry-wide level.