Anthropic has accused Chinese tech giant Alibaba of orchestrating what it describes as the largest known distillation attack of its kind, allegedly using tens of thousands of fraudulent accounts to systematically extract capabilities from its AI model, Claude. The accusations were reportedly made directly to US lawmakers and signal a significant escalation in tensions over AI intellectual property between American and Chinese technology firms.
The alleged operation reportedly involved nearly 25,000 fake accounts and generated close to 29 million exchanges with Claude over just six weeks, a volume that underscores how industrialized this type of AI extraction activity may be becoming.
The claims add Alibaba to a growing list of Chinese AI developers that Anthropic says have targeted its models. They also arrive at a time when the global race for AI supremacy is intensifying at a pace that is reshaping both corporate strategy and geopolitics.
Inside the Alleged Campaign
According to Anthropic’s account to US lawmakers, operators linked to Alibaba and the company’s Qwen AI team generated 28.8 million exchanges with Claude between April 22 and June 5, 2026, a roughly six-week period. The alleged campaign did not target Claude’s source code, model weights, or original training data. Instead, Anthropic claims the fraudulent accounts repeatedly posed carefully designed questions to Claude and collected the responses at scale.
That process is known as distillation, a technique in which a company uses the outputs of a larger, more capable AI model to train a smaller, cheaper one. The smaller model learns to replicate the useful behaviors of the more powerful system without requiring the same computing resources.
The areas targeted were notably specific. The alleged campaign focused on software development, multi-step reasoning, and agentic tasks, the kinds of capabilities that allow an AI model to plan and execute complex, multi-stage work rather than simply respond to individual prompts. Anthropic told lawmakers the campaign could help Chinese AI developers approach the capabilities of its Mythos Preview model, which is focused on advanced cybersecurity work, including identifying and exploiting complex software vulnerabilities.
An Industry Under Pressure
Francesco Pili, Counsel at Hogan Lovells, said the accusations are part of a pattern that extends well beyond Anthropic.
“This is part of a broader trend of regulatory disputes over whether, and on what terms, AI systems can access third-party platforms, services, or data,” he said.
“These cases are blurring the line between traditional platforms and agentic AI services. They are also testing the limits of what target platforms may lawfully do in response, including whether, and how, they can restrict access.”
Pili pointed to the dispute between Perplexity and Amazon as another example. He said it raises questions about whether an agentic AI system can be held responsible for accessing a third-party platform through user accounts without permission.
Distillation is not inherently illegal or unethical, but Anthropic argues that collecting another company’s model outputs without permission and at industrial scale crosses a clear line. The company has asked US lawmakers to penalize firms engaged in such practices.
Anthropic argues that, if left unchallenged, the practice acts as a “massive subsidy for our geopolitical competitors.” The US and China are engaged in an accelerating race to become the world’s dominant AI superpower. China’s DeepSeek shocked Western observers when it emerged with capabilities comparable to leading US models, reportedly at a fraction of the training cost. Several of America’s leading AI companies have since warned against domestic regulation on the grounds that it would hand Beijing a strategic advantage.
Anthropic has made similar accusations against Chinese companies before. Earlier this year, the company alleged that DeepSeek, Moonshot AI, and MiniMax collectively generated more than 16 million Claude exchanges through approximately 24,000 accounts. The alleged Alibaba campaign produced nearly twice as many exchanges in a comparable timeframe.
What Comes Next
Anthropic’s decision to bring these accusations directly to US lawmakers rather than pursue immediate legal action suggests the company is seeking a policy response as much as a legal one. The scale of the alleged operation makes it difficult to dismiss as opportunistic or accidental.
As legal and regulatory frameworks governing AI access, data use, and model training continue to take shape, cases like this will likely help define where the boundaries are drawn. For now, the message from Anthropic is clear: the extraction of AI capabilities without consent is not a gray area, and the company intends to make that case at the highest levels of government.
