US artificial intelligence company Anthropic has accused Chinese technology giant Alibaba of carrying out the largest known attempt to illicitly extract capabilities from its Claude AI models, escalating concerns in Washington over the transfer of advanced American AI technology to China.

The campaign was conducted between April 22 and June 5, 2026, and generated more than 28.8 million exchanges with Claude through nearly 25,000 fraudulent accounts, Anthropic said in a letter seen by Reuters.

The company described the operation as a “distillation” effort. Model distillation typically involves using the outputs of a more capable AI system to train or improve a smaller or less advanced model.

Anthropic alleged that the campaign was carried out by operators affiliated with Alibaba and Alibaba Qwen, the Chinese group’s artificial intelligence laboratory.  

Must read: Anthropic’s Claude Fable 5 ban sparks lawsuit against US Commerce Department

According to Anthropic, the operation could help accelerate China’s efforts to reproduce capabilities found in its advanced Mythos Preview model.

The June 10 letter was sent to US Senators Tim Scott and Elizabeth Warren, the chair and ranking member of the Senate Banking Committee, respectively, ahead of a scheduled hearing on artificial intelligence.

The allegations come amid heightened scrutiny in Washington over China’s access to advanced US artificial intelligence models and technologies. In April, the White House accused China of stealing intellectual property from American AI laboratories on an industrial scale.

Anthropic said it supported US government efforts to counter such operations, including greater threat-intelligence sharing between the government and private-sector AI companies.

The Alibaba campaign, if confirmed, would be significantly larger than earlier distillation operations identified by Anthropic.

In a February post, the company said it had uncovered attempts by Chinese AI startup DeepSeek and two other AI laboratories to extract capabilities from Claude. Anthropic said DeepSeek’s operation involved more than 150,000 exchanges, while Moonshot AI generated over 3.4 million exchanges and MiniMax more than 13 million.

At the time, Anthropic warned that such campaigns were growing in “intensity and sophistication” and said tackling the threat would require “rapid, coordinated action among industry players, policymakers and the global AI community.”

Must read: Microsoft’s Satya Nadella calls for an AI reset; Warns against OpenAI-Anthropic dominance

The latest accusations also come as Alibaba faces additional regulatory pressure in the United States. The company was added this month to the Pentagon’s list of Chinese military companies, a designation that Alibaba is challenging.

The US Commerce Department has, however, held off placing DeepSeek on a trade blacklist despite an interagency government committee reportedly determining that the company posed a national security risk. The decision was aimed at avoiding a further escalation in tensions with Beijing.

Separately, on June 12, two days after Anthropic sent the letter, the Commerce Department imposed restrictions on Anthropic’s latest Mythos and Fable models over concerns that they could be used by military and intelligence organisations in China and other countries of concern.

The restrictions prompted Anthropic to disable access to the models globally, highlighting the growing tension between US efforts to maintain technological leadership and the commercial ambitions of its leading AI companies.