Anthropic Accuses Alibaba of Using 25,000 Fake Accounts to Distill Claude Model; U.S. Congress Eyes Sanctions

According to Bloomberg on Wednesday, Anthropic accused Alibaba's Qwen team of using approximately 25,000 fake accounts to make 28.8 million requests to Claude between April and June, targeting core capabilities including software engineering and agentic reasoning through distillation attacks. Anthropic called this the largest-scale model distillation attack on a U.S. AI system by a Chinese company, claiming Alibaba sought to replicate advanced AI technology at lower cost without authorization.

Alibaba's U.S. ADRs fell 2.73% to $99.80 on the news. Meanwhile, U.S. lawmakers are drafting legislation to authorize sanctions against Chinese firms using unauthorized outputs from American AI models for competitor development. The measures may be included in annual defense authorization bills.