Anthropic accused Chinese tech giant Alibaba of conducting a large-scale campaign to extract the capabilities of its Claude AI models, according to CNBC.
In a letter dated June 10 addressed to Sen. Tim Scott and Sen. Elizabeth Warren, Anthropic described the effort as "the largest known distillation attack on Anthropic to date." The letter was sent to the Senate Committee on Banking, Housing, and Urban Affairs.
The technique involves training a weaker model on the outputs generated by a more powerful one, allowing developers to replicate capabilities without building from scratch. According to the letter, those affiliated with Alibaba and its AI lab used approximately 25,000 fraudulent accounts to run nearly 28.8 million exchanges with Anthropic's models over a roughly six-week window ending June 5. Claude is not available to entities in China.
Among the specific Claude capabilities the campaign sought to harvest, the letter named agentic reasoning, software engineering proficiency, and long-horizon task completion, according to The Wall Street Journal. Anthropic argued that distillation attacks are turning "billions of dollars in American investment and R&D into a massive subsidy for our geopolitical competitors."
"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 said in a statement. Alibaba did not respond to requests for comment from either outlet. Bloomberg first reported the letter's existence.
Beyond the accusations themselves, the letter urged Washington to pursue stricter export restrictions on semiconductors and to pass laws that would impose penalties on AI developers caught running distillation operations. Anthropic pointedly noted that Alibaba pressed forward even after a Trump White House memo in April had put foreign entities on notice about industrial-scale efforts to harvest U.S. AI systems — a warning Anthropic said Alibaba chose to ignore.
This is not the first time Anthropic has raised the alarm over distillation. The company had previously flagged the same type of activity in February, when it publicly named DeepSeek, Moonshot, and MiniMax as labs it believed were running large-scale distillation operations against its models.
The accusation comes as Anthropic is navigating a separate dispute with the Trump administration. The company's two most powerful models, Fable 5 and Mythos 5, were pulled offline after the administration issued an export control order citing national security concerns. Senior Anthropic staff traveled to Washington to meet with White House officials, and both sides said they are working to resolve the matter.
Alibaba, meanwhile, is also contending with its own U.S. government conflict. The company filed a federal lawsuit against the Pentagon seeking removal from a list of companies with alleged ties to China's military, a designation it called "arbitrary and capricious" with "no basis in fact or law."