Anthropic reveals large-scale theft of Claude technology: operators linked to Alibaba carried out an attack using 25,000 fake accounts
Anthropic, the developer of the advanced AI model Claude, has identified the largest distillation attack in its history, orchestrated by operators linked to China's Alibaba and its Qwen research lab. This involves the systematic extraction of Claude's capabilities through fake accounts, posing a direct threat to the intellectual property and competitive advantages of American developers.
According to data gathered during the investigation, from April 22 to June 5, attackers created nearly 25,000 fake accounts, through which over 28.8 million interactions with the model were made. This scale indicates a well-planned and resource-intensive campaign aimed at obtaining key technical characteristics of Claude.
Anthropic emphasizes that the attack was notable not only for its scale but also for its "brazen nature." The company notes that Alibaba is listed on the New York Stock Exchange, conducts business in the US, and is accountable to American investors and regulators. This makes the situation particularly egregious, as the violation comes from an entity formally integrated into the legal framework of the United States.
Distillation is a method where a less powerful model is trained on the responses of a stronger one. In itself, it is often used legitimately, for example, to create compact versions of AI systems. However, the problem arises when competitors gain access to an advanced tool through fake accounts, bypass service restrictions, and use the responses to train their own systems without the costs of development and training.
Attack Objectives and Economic Consequences
According to Anthropic, the alleged campaign targeted Claude's capabilities in agentic tasks, software development, and long-term planning. The developer stresses that such actions allow replicating the behavior of a frontier model without the cost of training it, which "upends the economic logic underpinning US leadership in AI." In effect, billions of dollars in American investments are turned into a subsidy for competitors.
Furthermore, unauthorized distillation could accelerate the development of Chinese AI systems for cyber operations, military tasks, and intelligence, posing a direct threat to national security.
What Anthropic Demands from Congress
In a letter to lawmakers, Anthropic calls for expanding the sharing of technical indicators and intelligence between frontier AI model developers and the US government. The company also asks for clarification of antitrust rules so that firms can share information about such attacks without risking violations of competition law.
A separate set of proposals concerns export controls. Anthropic demands tightening restrictions on advanced AI chips and computing resources, as well as closing loopholes that allow Chinese organizations to access foreign data centers. The company proposes imposing sanctions or other measures against parties responsible for large-scale extraction of model capabilities.
Context and Previous Incidents
This is not the first time Anthropic has faced large-scale attacks from Chinese labs. In February, the company accused DeepSeek, Moonshot AI, and MiniMax of generating over 16 million interactions with Claude through approximately 24,000 fake accounts. At the time, developers linked the campaigns to specific labs via IP addresses, request metadata, and infrastructure indicators.
The incident with Alibaba occurs against the backdrop of tightening US legislation. In April, Congressman Bill Huizenga introduced a bill against the extraction of key technical characteristics of closed US AI models by foreign adversaries, providing for export restrictions and sanctions. Recall that in December 2025, Nvidia also reported developing technology to verify the location of its processors amid rumors of accelerator smuggling to China.
My Expert Assessment: This incident demonstrates that the battle for technological leadership in AI is entering a new phase. Large Chinese corporations, acting through shell structures, are systematically extracting the intellectual property of American developers. This is not just a violation of terms of service—it is a strategic threat requiring immediate and tough countermeasures from both business and the state. The question is not whether sanctions will be imposed, but how quickly and effectively they can close existing loopholes.