Anthropic accused Alibaba of waging a large-scale effort to “illicitly” access its Claude AI model through almost 25,000 fraudulent accounts and 28.8 million exchanges between April and June, in a letter to US senators and White House officials reported by Bloomberg. The company said operators linked to Alibaba’s Qwen lab targeted Claude’s most prized capabilities like software engineering and agentic reasoning, calling it the biggest attempt yet by a Chinese firm to piggyback on top US labs through “adversarial distillation,” where one model is trained on another’s outputs to replicate it at a fraction of the cost. Alibaba’s US-listed shares fell more than 3% on the news, and the company had no comment. The accusation feeds a broader Washington push, with bipartisan lawmakers moving to blacklist or sanction Chinese firms found improperly accessing US AI output.
