China Conducts Industrial Distillation Of U.S. Models

U.S. cybersecurity and AI officials say Chinese labs DeepSeek, Moonshot AI, and MiniMax conducted industrial-scale "adversarial distillation" against OpenAI, Anthropic, and Google models in early 2026, using over 24,000 fraudulent accounts and more than 16 million exchanges. Analysts say the extraction undermines U.S. compute advantages, raises safety and national-security risks, and recommends layered U.S. legal measures including BIS entity listings and sanctions.