AI · Policy · 2026-09-09 13:56:00
The NSA, CISA, and FBI jointly accused six Chinese AI firms—DeepSeek, Moonshot AI, Alibaba, MiniMax, StepFun, and Z.AI—of industrial-scale distillation attacks on US frontier models since late 2024, likely with Chinese government awareness. Attack methods include bulk-buying fake accounts to exploit inference APIs and prompt injection to jailbreak models. Agencies recommended improved detection, identity verification, and subtly degrading responses or secretly switching malicious accounts to inferior models. China's Foreign Ministry rejected the accusations as groundless.
Auto-generated summary for reference only. RSS