Frontier-Model Distillation and Export Controls Draw Government Scrutiny

A September 8 joint NSA, CISA, and FBI advisory alleges China-based AI firms systematically extracted capabilities from major U.S. frontier models using fraudulent accounts, APIs, cloud services, proxies, prompt injection, and agentic workloads; adaptive attacks reportedly defeated all ten tested fingerprinting schemes. A reported DHS order directs federal agencies to remove DeepSeek products and exclude them from procurement, reportedly extending to contractors; scope and technical rationales remain uncertain. UK funding totals remain unresolved.