In a confidential letter to US senators, the AI safety company alleges nearly 25,000 fake accounts generated over 28.8 million unauthorised exchanges with Claude in just six weeks
Anthropic has filed a complaint against Alibaba accusing it of orchestrating a sophisticated and unprecedented campaign to clone its Claude AI model. The excerpts from a letter written by Anthropic and addressed to U.S. Senators Tim Scott and Elizabeth Warren demonstrate some of the current tensions between the U.S. and China regarding the AI technologies.
The Scale of the Operation
According to Anthropic, operators linked to Alibaba and its AI division Qwen used approximately 25,000 fraudulent accounts to gain unauthorised access to Claude between April 22 and June 5, 2026 — a window of just six weeks. During that period, those accounts generated over 28.8 million exchanges with the chatbot, all allegedly in violation of Anthropic’s terms of service.
What They Were After
Anthropic contends the campaign was not random scraping but a targeted extraction effort. The alleged objective was to replicate Claude’s most sophisticated and commercially valuable capabilities, specifically its agentic reasoning, software engineering competencies, and long-horizon task execution abilities — precisely the features that distinguish frontier AI models from their competitors.
What Anthropic Is Asking For
The complaint is as much a policy intervention as a legal one. Anthropic is urging US lawmakers to introduce stronger legal protections against AI model distillation — the technical process of extracting capabilities from one AI system to train another. The company is calling for tougher penalties targeting Chinese AI firms engaged in such practices, framing the issue as a matter of national technological security rather than a routine intellectual property dispute.
Why It Matters
The implications of this incident raise concerns over how easily frontier AI models can be extracted systematically on a mass scale. As AI becomes one of the most valuable assets for companies and for national-security purposes, the case between Anthropic and Alibaba may provide the impetus for governments to consider AI model misappropriation as severely as they do semiconductor espionage.
Senate hearings regarding artificial intelligence will likely explore these topics further, which could form a basis for a new legislative framework to govern AI protection and cross-border technology protections.
