US Agencies accuse six Chinese AI firms of industrial-scale frontier-model distillation
US agencies (NSA, CISA, FBI) accuse six China-based AI firms—DeepSeek, Moonshot AI, Alibaba, MiniMax, StepFun, and Z.AI—of conducting industrial-scale distillation to extract capabilities from US frontier models (Claude, GPT, Gemini, Grok) since 2024, using tactics like bulk fake accounts and prompt-injection. They urge mitigations such as stricter identity verification, anomaly monitoring, downgrading suspected outputs, and secretly switching users to an inferior model, while calling for information-sharing to track distillation trends. China denies the claims, framing them as groundless and arguing for greater AI cooperation. The case highlights escalating US-China AI tensions and potential policy shifts to safeguard US-led model capabilities.
- Six Chinese AI firms accused of aggressively copying US frontier models Ars Technica
- China ‘Willing’ to Hold AI Talks With US Despite Row Over Models Bloomberg.com
- US claims Chinese AI firms are carrying out ‘industrial-scale’ theft of trade secrets CNN
- US accuses Chinese AI firms of 'malicious' copying of AI technology Reuters
- Is China Stealing American AI? Why ‘Distillation’ Has Washington Up in Arms WSJ
Reading Insights
1
7
8 min
vs 9 min read
94%
1,721 → 100 words
Want the full story? Read the original article
Read on Ars Technica