China and US on collision course over AI training, funding disagreements

The article outlines rising U.S.–China tensions around AI model “distillation,” with a White House memo accusing China-based actors of running “industrial-scale” campaigns to copy U.S. frontier models and proposing four defensive and enforcement measures. Beijing has pushed back and is reportedly preparing rules to restrict Chinese tech firms from accepting U.S. investment without approval after Meta’s roughly $2 billion Manus acquisition triggered scrutiny. Separately, Hong Kong-based DeepSeek previewed its V4 model, claiming near-parity with top-tier models at a fraction of the cost (2 yuan vs. 216 yuan per 1M tokens for GPT-5.5). Taken together, regulatory frictions and cheaper competitive Chinese models add geopolitical and competitive risk to U.S. AI/tech stocks (notably META.OQ) and could pressure valuations and cross-border deal activity.