The Politics of AI Procurement

Design a framework for how a government should evaluate AI systems for public administration or intelligence work. Address data sensitivity, vendor lock-in, auditability, legal jurisdiction, model performance, costs, and the difference between buying a product and building sovereign capability. Include criteria for when a state should prefer open-source, local vendors, or major global suppliers, and explain how to avoid replacing one dependency with another.