Alibaba Allegedly Steals AI Knowledge from Claude

Anthropic, the maker of the Claude chatbot, claims that between April 22 and June 5, 2026, a group linked to Alibaba used fake accounts—about 25, 000 in total—to chat with Claude over 28. 8 million times. The company says this was an effort to copy Claude’s abilities so that Alibaba could build a comparable model faster. The technique, known as distillation, lets one system learn from another by mimicking its responses. Anthropic warns that this is a common way for rivals to steal U. S. AI technology and help foreign firms catch up quickly.A letter dated June 10, sent to U. S. Senators Tim Scott and Elizabeth Warren before a Senate Banking Committee hearing on AI, details the alleged scheme. It names operators tied to Alibaba’s AI lab, Qwen, as the perpetrators. Alibaba has not yet replied to a request for comment. If Anthropic’s claims are true, the move could shift the balance of power in AI research and raise questions about how companies protect proprietary models. The incident also highlights the growing tension between U. S. and Chinese tech firms over data, innovation, and national security.