Anthropic Says Seven China-Based AI Labs Ran Industrial-Scale Claude Distillation Attacks

Summary: Anthropic said it identified and disrupted seven industrial-scale illicit distillation campaigns against Claude run by China-based labs, including Alibaba, Moonshot, DeepSeek, Z.ai (Zhipu), and MiniMax, since February 2026. The labs gained access through proxy services that created thousands of accounts using fictitious identities, fake or stolen credit cards, and harvested API keys, and Anthropic said DeepSeek, Xiaomi, and Moonshot fed conversations between their own models and users into Claude and used its responses as training data, with some exchanges containing sensitive information from individual users, major multinational companies, and state-affiliated actors.
What you should know: Anthropic says proxy networks both serve users in unsupported regions and save their exchanges to sell to other labs, so conversations your employees have with a frontier model can become another company's training data without their knowledge. The same week, U.S. cybersecurity and intelligence agencies accused China-based AI companies of systematic extraction of American frontier models.

Read the original at The Hacker News ↗