U.S. artificial intelligence company Anthropic has publicly accused Chinese tech giant Alibaba of illicitly accessing its Claude AI models in what it calls the largest known "cloning" attempt against its systems. According to Ars Technica, Anthropic alleges Alibaba used roughly 25,000 fake accounts to mine Claude across 28.8 million exchanges over a 44-day period. PYMNTS reported a figure of 29 million queries. Several outlets, including the Indian Express, tie the effort to Alibaba's Qwen AI team.
The technique at issue is known as "distillation," in which one AI system is fed enormous numbers of queries to extract the patterns and capabilities of another model and effectively copy them. Anthropic frames the activity as theft of its AI capabilities rather than ordinary use.
Anthropic has taken the dispute to Washington. According to Benzinga, the company wrote to U.S. lawmakers including Senators Elizabeth Warren and Tim Scott, and other reports say it notified the White House, urging tougher penalties on Chinese AI firms.
The accusation hit Alibaba's stock. Bloomberg, via Techmeme, reported Alibaba shares fell about 5% in Hong Kong to a 16-month low, deepening a year-to-date decline of 33%, with Xiaomi and Baidu also falling more than 3%. One outlet, TechStock², put the wipeout at roughly HK$88 billion in market value.
The sources presented here are Anthropic's allegations; they do not include a detailed response or rebuttal from Alibaba, and the claims have not been independently confirmed.
Why it matters: the case sharpens an escalating U.S.–China fight over who owns AI capabilities, and could push Washington toward stricter rules on how foreign firms access American models.