The NSA, CISA and FBI issued a joint cybersecurity advisory accusing Chinese AI companies of "distillation activities at an industrial scale." The agencies allege DeepSeek, Moonshot AI, Alibaba, MiniMax, StepFun and Z.AI extracted billions of tokens since 2024 from models including Claude, GPT, Gemini and Grok. The advisory says DeepSeek used that data to train R1, while Moonshot AI used Claude's Fable to train Kimi K3. It also lists mitigations US firms can implement against such campaigns.
No score is assigned. Sources and their independence are shown in the citation chain below.