Chinese AI Firms Hit Claude with Distillation Attacks, Anthropic Warns

Anthropic revealed that three China-based AI labs conducted large-scale model distillation attacks against its Claude LLM—generating over 16 million exchanges via about 24,000 fraudulent accounts—to illicitly copy model capabilities; Anthropic attributed the activity using IP and metadata correlation and has deployed detection systems, chain-of-thought and coordinated-account detection, stronger verification for high-risk accounts, and product/API safeguards to mitigate further misuse and security risks.