US agencies allege Chinese AI firms steal US model capabilities; Anthropic claims corroborate

US intelligence agencies and Anthropic allege Chinese AI developers systematically extract capabilities from US frontier models to reduce compute and energy costs. The claim is based on a joint NSA/FBI/CISA advisory and Anthropic's assertions, but lacks public technical evidence. If true, it signals a significant escalation in US-China tech competition and could accelerate export controls and AI security measures.