Anthropic has accused Alibaba and its Qwen AI team of carrying out a large-scale “distillation attack” on its Claude model. The company alleges that proprietary capabilities were extracted and knowledge was transferred in the process. According to Anthropic, nearly 25,000 fake accounts were used to generate around 28.8 million interactions with Claude between April and June 2026.
The goal, it claims, was to replicate advanced capabilities such as reasoning, coding, and long-horizon task execution in competing models. Anthropic says this violates its terms of service and represents the largest known attempt to copy its model’s capabilities. Alibaba has not publicly confirmed the allegations, which are now drawing attention from U.S. regulators.