Anthropic Accuses Alibaba of AI Model Misappropriation

U.S. AI company Anthropic has accused Alibaba of illicitly extracting its AI model capabilities in what it claims is the largest known attack on the company. The campaign, linked to Alibaba and its AI lab, aimed to accelerate China’s AI progress and involved millions of interactions with illegitimate accounts.

U.S.-based artificial intelligence firm Anthropic has lodged accusations against Chinese tech giant Alibaba for allegedly misappropriating its Claude AI model capabilities. According to an exclusive letter reviewed by Reuters, this marks the most significant breach of its kind against Anthropic.

The attack entailed a 'distillation' process, allegedly orchestrated by parties connected to Alibaba and its AI lab, Alibaba Qwen. Anthropic stated that the operation spanned from April 22 to June 5, 2026, involving over 28.8 million interactions through nearly 25,000 fraudulent accounts intent on bolstering China's AI offerings.

This accusation follows precedents of similar threats from other Chinese AI labs, such as DeepSeek, prompting calls for urgent collaborative action among governments and the AI sector. The situation adds tension to U.S.-China tech relations amid the U.S. government’s ongoing concerns about intellectual property theft.