Story perspectives
Anthropic Alleges 28.8 M Fake AI Outputs; Admin Flags Threat
7/14/2026
1 of 3
Anthropic Accuses
- Anthropic sent a letter to Senators Tim Scott and Elizabeth Warren accusing Alibaba, DeepSeek, Moonshot and MiniMax of distillation that generated 28.8 million Claude outputs using 25,000 fake accounts.
- Z.ai released GLM-5.2 last month, a model said to match U.S. coding and cybersecurity.
- The Trump administration called distillation a security threat while experts say a crackdown would have limited impact because it cannot build GLM-5.2, even as Congress debates the AI Overwatch Act.
1 / 3
