Executive Summary
Anthropic said it disrupted industrial-scale illicit distillation attacks against Claude from seven China-based labs, including Alibaba, Moonshot, DeepSeek, Z.ai and MiniMax.
What Happened
Anthropic on Thursday said it identified and disrupted industrial-scale illicit distillation attacks against Claude from seven labs based in China, including Alibaba, Moonshot, DeepSeek, Z.ai (aka Zhipu), and MiniMax. Knowledge distillation by itself is a legitimate training method. It refers to a machine learning technique where a large, powerful AI model assumes the role of a "teacher" to
Affected Products
- Anthropic Claude
Vulnerability / Attack Technique
Model distillation through large-scale querying
Technical Mechanism
The labs queried Claude at scale to use its outputs as teacher data for training their own models.
AI Security Category
LLM10:2025 Unbounded Consumption
Relevant Frameworks
RealGround Analysis
Model capability is being extracted through ordinary API access. RealGround analysis: companies exposing fine-tuned or proprietary models via API face the same risk at smaller scale.
Recommended Controls
- Monitor API usage for distillation patterns
- Apply per-account and cross-account rate limits
Related RealGround Service
AI Red Teaming · or start with the free AI Agent Security Scanner.
Sources
thehackernews.com: https://thehackernews.com/2026/09/anthropic-says-seven-china-based-ai.html
Summary and analysis written by RealGround from the source above; last updated 2026-10-17.