Anthropic claims Alibaba defied Trump to attack Claude and steal capabilities

Anthropic has made serious allegations against Chinese tech giant Alibaba, claiming that the company orchestrated an extensive campaign aimed at cloning its AI model, Claude. This claim comes as China intensifies its efforts to match the capabilities of leading AI models like Claude, particularly following the release of Mythos and subsequent limitations on access to foreign markets. In a letter dated June 10, addressed to Senators Tim Scott and Elizabeth Warren ahead of a Senate committee hearing focused on AI and its impact on the American economy, Anthropic revealed what it termed 'confidential evidence' of the most significant campaign to illicitly extract Claude's capabilities to date. The period of these alleged attacks spanned from April 22 to June 5, during which Anthropic stated that operators linked to Alibaba and its AI lab, Alibaba Qwen, engaged in over 28.8 million interactions with Claude through nearly 25,000 fake accounts. Anthropic claims that this operation violated Claude's terms of service and sought to exploit some of the model's most critical features, including agentic reasoning, software engineering, and long-term task management. The company further noted that Alibaba managed to evade detection by employing sophisticated obfuscation techniques and proxy networks. As demand for reliable obfuscation methods grows in China, Anthropic warned of a burgeoning circumvention economy that could lead to an increase in such distillation attacks in the future. The allegations point to a broader trend among Chinese labs attempting to replicate US frontier models, with Anthropic asserting that Alibaba's goal was to extract Claude's capabilities without incurring the substantial costs associated with training their own advanced models. Anthropic emphasized that these attacks are becoming increasingly prevalent, effectively turning billions of dollars in American investment and research into a significant advantage for geopolitical competitors. Notably, these allegations against Alibaba surfaced after former President Donald Trump took measures to combat similar illicit distillation attacks, which he described as 'industrial-scale' AI theft, following previous accusations against other Chinese firms like DeepSeek and Moonshot. Anthropic also noted that similar attacks on models from OpenAI and Google have been documented, underscoring the growing concern over AI security and intellectual property in the tech industry.