Anthropic Accuses Chinese AI Firms of Illicit Claude Model Exploitation

• Anthropic alleges three Chinese AI companies—DeepSeek, Moonshot AI, and MiniMax—conducted large-scale "distillation" campaigns targeting its Claude AI model.
• The firms reportedly created over 24,000 fake accounts and executed more than 16 million exchanges with Claude to extract advanced capabilities like agentic reasoning and coding.
• While AI model distillation is common, Anthropic argues that using a competitor’s proprietary system at this scale constitutes intellectual property infringement.
• DeepSeek, Moonshot AI, and MiniMax engaged in varying levels of interaction, with MiniMax diverting nearly half its traffic to Claude's latest features.
• Anthropic warns that models built through illicit extraction may lack safety guardrails, posing strategic risks beyond commercial concerns.