What happened
RAND published this research report on 11 September 2026 arguing that public procurement offers 'AI middle powers' — states with significant economic, diplomatic or regulatory capability but no domestic frontier AI firms — a practical mechanism to govern their dependency on foreign frontier AI providers. The report identifies three categories of risk that fall disproportionately on state buyers in these countries: vendor lock-in/service withdrawal threats, limited visibility into frontier AI development, and data sovereignty/information-security risks. It develops a framework of procurement obligations governments could negotiate with AI providers, including model access for evaluation, information sharing, funding for reliability research, state-of-the-art risk management requirements, and termination/exit provisions.
Why it matters
Government CIOs and procurement leads in countries without domestic frontier labs (most of the world) now have a concrete negotiating framework to reduce vendor lock-in and sovereignty risk when contracting with US or Chinese frontier AI providers.
Action needed
Government technology and procurement leadership in AI middle-power jurisdictions should benchmark current AI vendor contracts against the report's proposed procurement obligations (evaluation access, exit provisions, information sharing).