Forget Forty Thieves — Anthropic Says Alibaba Used 25,000 Fake Accounts to Steal Its AI
Anthropic sent a letter to Congress accusing Alibaba of running 28.8 million fake conversations with Claude to train a competing AI for free.
In the classic folk tale, Ali Baba outwitted the forty thieves. According to Anthropic, the Chinese tech giant Alibaba has been running an elaborate heist of its own, only with 25,000 fake accounts, reports Bloomberg.
In a letter sent to US senators and White House officials, Anthropic accused Alibaba of orchestrating the largest known AI theft campaign against the company to date. Between April 22 and June 5, operators linked to Alibaba’s Qwen AI lab ran 28.8 million exchanges with Claude through fraudulent accounts — targeting software engineering and agentic reasoning.
Anthropic alleges Alibaba pumped Claude full of questions, harvested the answers and used them to train a competing model at a fraction of the cost. Building a frontier AI model costs billions. Stealing one, apparently, just takes fake accounts and patience.
Alibaba declined to comment. Anthropic has since banned all accounts involved and is pushing Congress to sanction firms caught doing the same. Anthropic, OpenAI and Google have already joined forces to share intelligence on such attempts.
In the classic folk tale, Ali Baba outwitted the forty thieves. According to Anthropic, the Chinese tech giant Alibaba has been running an elaborate heist of its own, only with 25,000 fake accounts, reports Bloomberg.
In a letter sent to US senators and White House officials, Anthropic accused Alibaba of orchestrating the largest known AI theft campaign against the company to date. Between April 22 and June 5, operators linked to Alibaba’s Qwen AI lab ran 28.8 million exchanges with Claude through fraudulent accounts — targeting software engineering and agentic reasoning.
Anthropic alleges Alibaba pumped Claude full of questions, harvested the answers and used them to train a competing model at a fraction of the cost. Building a frontier AI model costs billions. Stealing one, apparently, just takes fake accounts and patience.
Alibaba declined to comment. Anthropic has since banned all accounts involved and is pushing Congress to sanction firms caught doing the same. Anthropic, OpenAI and Google have already joined forces to share intelligence on such attempts.
