Anthropic Claims Chinese AI Companies are “Distilling” Claude
Anthropic has accused three Chinese AI firms of misusing its Claude chatbot to strengthen their own systems, saying the activity violated its policies and highlights the need for tighter controls on advanced chip exports. In a recent statement, the San Francisco-based company said MiniMax, DeepSeek, and Moonshot generated more than 16 million combined exchanges with Claude through approximately 24,000 fraudulent accounts. The company said the firms relied on a method known as distillation, a practice in which a smaller or less advanced model is trained using responses produced by a more capable system. Anthropic warned that such efforts are becoming more organized and technically advanced. The company described the activity…