U.S. AI company Anthropic has accused Chinese technology giant Alibaba Group Holding Ltd. of conducting the largest known distillation attack on the company to date. In a letter sent to White House officials and U.S. Senate Banking Committee members Tim Scott and Elizabeth Warren, Anthropic alleged that operators affiliated with Alibaba and its AI lab, Alibaba Qwen, illicitly extracted the capabilities of the Claude AI model.
The distillation attack, which took place between April 22 and June 5, 2026, involved using nearly 25,000 fraudulent accounts to generate more than 28.8 million exchanges with Claude. According to Anthropic, distillation attacks are a method where AI companies improperly obtain capabilities by training a less capable model on the outputs of a stronger one.
Anthropic stated that this campaign is a way for competitors to harvest American AI and helps China reach advanced model capabilities sooner. The company further noted that the effort undermines its decision to keep its products out of China. Alibaba did not immediately respond to requests for comment.