Frontier-Model Distillation and Misuse Claims Escalate AI Security Competition

Anthropic alleges nearly 200 million exchanges across five model-distillation campaigns involving Alibaba, Moonshot AI, and DeepSeek, including apparent military-linked routing. The claims remain company-reported and require independent scrutiny, while calls for U.S. labs to distill frontier models intensify disputes over consent, copyright, model extraction, access controls, and geopolitical security.