Anthropic accuses Alibaba of massive AI ‘distillation’ attack

Anthropic has accused Chinese technology giant Alibaba of orchestrating what it describes as the largest known attempt to extract intelligence from its Claude AI models, escalating tensions in the global race for artificial intelligence dominance. The allegations have intensified concerns over intellectual property protection as AI companies battle to develop increasingly powerful large language models.

According to Anthropic, the alleged operation involved a sophisticated network of thousands of fraudulent user accounts that interacted with Claude over several weeks to systematically collect model outputs. The company claims the activity generated nearly 29 million conversations, making it the most extensive AI "distillation" campaign it has detected so far.

Anthropic alleges the operation was linked to Alibaba and its AI research division, Qwen, one of China's fastest-growing artificial intelligence projects. The company believes the collected responses were intended to improve competing AI systems without investing the enormous resources typically required to train frontier models from scratch.

Distillation is a widely used technique in artificial intelligence, allowing developers to train smaller models using outputs from more advanced systems. However, Anthropic argues that using another company's proprietary AI without authorization crosses the line from legitimate research into intellectual property theft.

The company says the campaign specifically targeted Claude's strongest capabilities, including advanced reasoning, complex software engineering tasks, and long-duration problem solving. These areas have become critical competitive advantages as AI companies race to build increasingly capable digital assistants and enterprise AI tools.

Anthropic has warned U.S. policymakers that such large-scale extraction efforts could enable foreign competitors to narrow the technological gap much faster than would otherwise be possible. The company argues that protecting advanced AI models has become a matter of both commercial competitiveness and national security.

The accusations arrive during an increasingly intense technology rivalry between the United States and China. Both countries are investing heavily in artificial intelligence, while governments continue tightening restrictions on advanced AI technologies, semiconductor exports, and high-performance computing infrastructure.

Anthropic has previously raised concerns about similar activities involving other Chinese AI developers, arguing that unauthorized model distillation is becoming a growing challenge across the industry. AI companies are now investing more heavily in monitoring systems designed to detect automated extraction attempts and abnormal user behavior.

Alibaba has not publicly responded to the allegations. If the claims are substantiated, the dispute could further complicate technology relations between Washington and Beijing and add momentum to ongoing debates surrounding AI governance, cybersecurity, and intellectual property enforcement.

As artificial intelligence becomes increasingly valuable, experts expect conflicts over model security, proprietary data, and unauthorized AI replication to become one of the defining legal and competitive battles shaping the next phase of the global AI industry.