Alibaba shares fell 3% after Anthropic accused the company’s Qwen AI lab of using 25,000 fraudulent accounts to scrape 28.8 million exchanges from its Claude AI model between April and June. Anthropic warns this industrial-scale adversarial distillation breaches its distribution policies and undermines AI safety guardrails.
Shares of Alibaba fell 3% after Anthropic sent a letter to U.S. officials accusing Alibaba’s Qwen AI lab of illicitly extracting capabilities from its Claude model. The market reaction reflects investor concern over potential legal, regulatory and reputational fallout.
Anthropic’s letter states that between April and June, operators linked to Alibaba launched a coordinated campaign using approximately 25,000 fraudulent accounts to conduct 28.8 million exchanges with the Claude AI model. This systematic approach allowed extraction of model outputs without incurring R&D or training costs.
Anthropic warns policymakers that such adversarial distillation practices compromise safety guardrails and violate distribution terms by harvesting U.S. AI capabilities at industrial scale. The startup urges stricter regulatory measures to deter unauthorized data scraping and protect intellectual property.
