Anthropic Accuses Alibaba of Large-Scale AI Technology Theft Through Fraudulent Access

Anthropic claims Alibaba conducted millions of illicit interactions with its AI models, accusing the Chinese tech giant of intellectual property theft via an AI distillation attack.

Editorial Team · Jun 26, 2026 · 2 min read · Source: digitalmusicnews.com

Anthropic, the AI firm behind the Claude chatbot, has formally accused Alibaba of executing an extensive and unauthorized effort to steal its artificial intelligence technology. The company detailed these allegations in a letter addressed to the U.S. Senate Committee on Banking, Housing, and Urban Affairs, describing Alibaba’s actions as “the largest known distillation attack” against Anthropic to date.

The accusation centers on a method known as AI distillation, which involves repeatedly querying a sophisticated AI model (“teacher”) through an API and harvesting the output to train a separate, competing AI system (“student”). According to Anthropic, Alibaba employees used nearly 25,000 fraudulent accounts to carry out approximately 28.8 million interactions with Anthropic's models over a period spanning from late April to early June.

This form of intellectual property theft raises concerns over the security of proprietary AI developments and highlights the need for coordinated defense strategies. Anthropic emphasized that tackling such illicit distillation requires combined efforts from both government and industry. The company pledged ongoing cooperation with U.S. lawmakers and federal agencies to safeguard American innovation in AI technologies.

The allegations come shortly after the White House Office of Science and Technology Policy issued a memorandum promising support for AI companies facing similar large-scale distillation threats. Earlier this year, Anthropic disclosed prior industrial-scale attacks linked to three smaller AI firms—DeepSeek, Moonshot, and MiniMax—which operated over 24,000 fraudulent accounts to generate more than 16 million exchanges with Claude's AI system.

Meanwhile, Anthropic is also defending itself amid separate lawsuits filed by major music publishers, including BMG, Concord, Universal Music Publishing Group, and ABKCO. These companies claim Anthropic unlawfully used copyrighted lyrics on a massive scale to train its AI models, alleging billions of dollars in damages. Anthropic, however, argues that training AI models qualifies as fair use under current copyright law.