Six Chinese AI firms accused of aggressively copying US frontier models

A joint advisory from the NSA, CISA, and the FBI has accused six Chinese artificial intelligence companies—DeepSeek, Moonshot AI, Alibaba, MiniMax, StepFun, and Z.AI—of systematically distilling capabilities from advanced US frontier models. According to the agencies, these firms have been targeting models such as Claude, GPT, Gemini, and Grok since late 2024. The report alleges that these activities, likely conducted with the awareness of the Chinese government, allow these companies to significantly accelerate their development timelines while bypassing the massive financial costs associated with training frontier-level AI from scratch. This industrial-scale distillation is framed as a major national security concern, as it undermines the competitive advantage of US-based AI developers. The agencies emphasize that this strategy enables Chinese firms to achieve performance parity with top-tier American models by leveraging stolen intellectual property rather than original research and development.