Anthropic finds 200M distillation attacks from Alibaba, Moonshot AI, DeepSeek
anthropic detected nearly 200 million exchanges linked to distillation attacks from china-based ai companies, targeting claude's capabilities. this affects us frontier models.
anthropic detailed distillation campaigns from alibaba, moonshot ai, and deepseek, which targeted claude's capabilities like agentic capabilities, tool use, coding, and data analysis. nearly 200 million exchanges were linked to these attacks, with alibaba's campaign being the largest. these attacks involved sophisticated methods to circumvent anthropic's defenses and harvest model capabilities. this raises concerns about model security and potential misuse.