Particle.news
Download on the App Store

Anthropic Accuses Three Chinese AI Labs of Industrial-Scale Claude Distillation

Anthropic urges a coordinated response to curb capability transfer through covert model‑extraction.

Overview

  • Anthropic alleges DeepSeek, Moonshot AI and MiniMax used roughly 24,000 fraudulent accounts to generate more than 16 million exchanges with Claude to train rival systems.
  • MiniMax is accused of running the largest operation with over 13 million exchanges, with campaigns targeting advanced reasoning, coding and tool use.
  • The company says it attributed the activity using IP correlations, request metadata and infrastructure signals, and that the labs routed traffic through commercial proxy services to evade geofencing.
  • Anthropic reports it has disrupted known access points, deployed behavioral fingerprinting, and shared indicators with U.S. government partners, while noting no evidence of direct Chinese government coordination.
  • OpenAI and Google have recently reported similar model‑extraction attempts, the named firms had not provided substantive responses in the covered reporting, and Anthropic has not announced lawsuits.