Anthropic Accuses Alibaba of Large-Scale AI Model Distillation Campaign
U.S. artificial intelligence company Anthropic has accused Chinese technology and e-commerce giant Alibaba of carrying out what it described as the largest known attempt to extract the capabilities of its Claude AI model through illicit means.
According to a letter seen by Reuters, Anthropic alleged that the campaign relied on a technique known as “distillation”. The company has previously described distillation as a process in which a less capable AI model learns from the outputs of a more advanced model.
Anthropic said the activity took place between April 22 and June 5, 2026. During that period, the campaign generated more than 28.8 million exchanges with Claude through nearly 25,000 fraudulent accounts.
Anthropic Details the Alleged Campaign
Anthropic stated that the campaign aimed to accelerate China’s ability to match the advanced capabilities available in its Mythos Preview models.
Furthermore, the company alleged that operators affiliated with Alibaba and Alibaba Qwen, the company’s AI research lab, conducted the activity. Alibaba did not immediately respond to a request for comment.
The letter, dated June 10, was addressed to Senators Tim Scott and Elizabeth Warren, the chair and ranking member of the U.S. Senate Banking Committee, before a scheduled hearing on artificial intelligence.
Previous Concerns Over AI Capability Extraction
Earlier, in April, the White House accused China of stealing intellectual property from U.S. AI laboratories on an industrial scale.
Anthropic also said it supports the U.S. government’s efforts to counter such attacks. These efforts include threat intelligence sharing and broader cooperation between government agencies and private AI companies.
Moreover, the company pointed to concerns it had already raised in February. At that time, Anthropic said it had identified campaigns by Chinese AI startup DeepSeek and two other Chinese AI laboratories that allegedly attempted to extract capabilities from its Claude AI platform.
According to Anthropic, the DeepSeek campaign involved more than 150,000 exchanges. It also claimed that Moonshot AI generated more than 3.4 million exchanges, while MiniMax exceeded 13 million.
Anthropic added that these campaigns had continued to grow in both intensity and sophistication. As a result, it argued that addressing the threat would require rapid and coordinated action among industry participants, policymakers and the wider AI community.
Broader U.S. Measures Affecting Chinese AI
Separately, Alibaba was added to the Pentagon’s list of Chinese military companies this month. The company is challenging that designation.
Meanwhile, the Commerce Department has delayed placing DeepSeek on a trade blacklist. Reuters exclusively reported this month that the department held back despite an interagency government committee identifying the company as a national security risk, as officials sought to avoid escalating tensions with Beijing.
In addition, on June 12, two days after Anthropic sent its letter, the Commerce Department imposed controversial restrictions on Anthropic’s latest Mythos and Fable AI models. Officials reportedly feared the models could be used by military intelligence users in China and other countries of concern.
Consequently, Anthropic disabled global access to both models following the restrictions.
With inputs from Reuters