Anthropic Accuses 3 Chinese AI Firms of 16M Claude Exchanges
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Anthropic accused DeepSeek, Moonshot, and MiniMax of creating roughly 24,000 fraudulent accounts and making more than 16 million exchanges with Claude in what it called 'industrial-scale campaigns' to distill its AI model.
- DeepSeek allegedly held over 150,000 exchanges with Claude, specifically targeting its reasoning capabilities and generating 'censorship-safe alternatives to politically sensitive questions about dissidents, party leaders, or authoritarianism,' per Anthropic.
- Moonshot and MiniMax were accused of more than 3.4 million and 13 million exchanges with Claude, respectively, as part of the same alleged distillation effort.
- Anthropic argued that illicitly distilled models are 'unlikely' to carry over existing safeguards and could be fed into 'military, intelligence, and surveillance systems — enabling authoritarian governments to deploy frontier AI for offensive cyber operations, disinformation campaigns, and mass surveillance.'
- OpenAI made a similar accusation in a letter to lawmakers last week, claiming DeepSeek was engaged in 'ongoing efforts to free-ride on the capabilities developed by OpenAI and other U.S. frontier labs.'
- Anthropic is calling on AI industry members, cloud providers, and lawmakers to address distillation, suggesting that 'restricted chip access' could limit both model training and the scale of illicit distillation.
Why it matters: Anthropic is publicly lobbying lawmakers, cloud providers, and the AI industry to treat Chinese distillation as a national security threat, explicitly tying the issue to 'restricted chip access.' OpenAI made a near-identical accusation against DeepSeek last week, signaling US frontier labs are aligning on a coordinated policy push that could shape export controls and chip restrictions targeting Chinese AI development.
Ask SkimNews
