Key facts
- Anthropic alleges Chinese AI firms Moonshot and DeepSeek engaged in large-scale model distillation.
- The firms allegedly routed user requests to Anthropic's Claude model to replicate its capabilities.
- Anthropic claims this practice is an attempt to distill and steal its proprietary AI technology.
- The company has observed an escalation in such attacks originating from China.
AI safety firm Anthropic has accused Chinese AI companies Moonshot and DeepSeek of engaging in large-scale model distillation. The companies allegedly routed user requests to Anthropic's Claude model to distill its advanced capabilities, a practice Anthropic views as an attempt to replicate its proprietary technology without authorization. Anthropic reported an escalation in these attacks originating from China, targeting Claude's sophisticated features.