Anthropic is accusing China's Alibaba of exploiting its AI models in a large-scale attack (www.businessinsider.com)

Anthropic has accused Alibaba, a major Chinese tech company, of conducting a large-scale "distillation attack" to illicitly extract capabilities from its advanced AI model, Claude. In a letter to U.S. Senators Tim Scott and Elizabeth Warren, Anthropic's head of policy, Sarah Heck, detailed how operators affiliated with Alibaba initiated nearly 29 million interactions with their model, utilizing approximately 25,000 fraudulent accounts between April and June. These distillation attacks, which involve leveraging advanced AI to train less sophisticated models, could enable Alibaba's AI systems to achieve advanced functionalities similar to those of Anthropic's cutting-edge models much sooner, without incurring significant research and development costs. This incident highlights a growing concern within the AI/ML community regarding intellectual property and the ethics of AI model training. Heck's call for stricter legislation aims to prevent such activities by limiting Chinese access to advanced U.S. computing infrastructure and imposing penalties on entities that engage in distillation attacks. The controversy follows recent U.S. export controls on Anthropic's Fable 5 model due to national security fears linked to foreign access to sensitive AI technologies. Additionally, this situation compounds pressure on Alibaba, which is facing scrutiny amidst a Pentagon blacklist designation as a business connected to the Chinese military.