Anthropic accuses Alibaba of using fraudulent accounts to harvest 28.8 million exchanges from its Claude AI model.
Anthropic accused Alibaba of deploying 25,000 fake accounts to extract 28.8 million exchanges from its Claude AI models between April and June 2026. The allegation, framed as a national-security concern, was detailed in letters to U.S. Senators Elizabeth Warren and Tim Scott.
Alibaba’s stock, already down 31% year-to-date, extended losses following the accusation. The dispute centers on claims that Alibaba used the harvested data to train its Qwen AI model, which has surpassed 1 billion downloads. Anthropic described the incident as the largest known distillation attack on its systems.
Shares reacted swiftly to the news, with investors pricing in regulatory and reputational risks. The accusation escalates tensions in the AI sector, where intellectual property disputes are increasingly tied to broader geopolitical concerns.