Anthropic has accused Chinese technology giant Alibaba of carrying out what it described as the largest known attempt to extract its artificial intelligence (AI) capabilities, according to a letter sent to the US Senate Committee on Banking, Housing, and Urban Affairs.
The letter, addressed to Senator Tim Scott and Senator Elizabeth Warren on June 10, alleged that Alibaba conducted “the largest known distillation attack on Anthropic to date”.
Distillation refers to a technique where a less advanced AI model is trained using the outputs of a more capable model. Anthropic claims the effort was aimed at reproducing the capabilities of its advanced Mythos Preview model.
The company said the campaign ran between April 22 and June 5, 2026, generating more than 28.8 million interactions with its Claude chatbot through nearly 25,000 fraudulent accounts.
Anthropic alleged that the operation was carried out by individuals linked to Alibaba and Alibaba Qwen, the company's AI research division.
“We believe combating the threat of illicit distillation requires coordinated action between government and industry, and we will continue working with Congress and the Administration to maintain American AI leadership,” an Anthropic spokesperson told CNBC.
The allegations come amid growing concerns among US AI companies about foreign competitors using their systems to improve rival models.
In February, Anthropic said it had uncovered similar efforts by Chinese AI startup DeepSeek and two other Chinese AI labs. According to the company, DeepSeek generated more than 150,000 exchanges with Claude, while Moonshot AI and MiniMax were responsible for over 3.4 million and 13 million exchanges, respectively.
Anthropic said at the time that such campaigns were becoming increasingly sophisticated and would require a coordinated response from industry, policymakers and the wider AI community.
According to a Wall Street Journal report, DeepSeek previously disclosed in a research paper that it trained its models using webpages and ebooks. Some of those webpages contained AI-generated responses, meaning the model may have indirectly learned from more advanced AI systems through publicly available internet data.
The dispute comes at a sensitive time for both companies. Earlier this month, Alibaba was added to the Pentagon's list of Chinese military companies, a designation it is contesting.
At the same time, the US Commerce Department imposed restrictions on Anthropic's latest Mythos and Fable AI models over concerns they could be used by military intelligence entities in China and other countries of concern. Anthropic subsequently disabled access to the models globally.
