Agencies Are Missing a Step to Share Information on Better AI Acquisition, GAO Finds

Why It Matters

Without a coordinated knowledge‑sharing framework, agencies risk repeating costly procurement errors and lag behind the rapid expansion of AI across government functions. Standardizing lesson‑capture will improve efficiency, reduce waste, and accelerate responsible AI deployment.

Key Takeaways

• AI use cases across agencies doubled from 2023 to 2024
• Agencies lack systematic collection of AI acquisition lessons
• OMB memo M-25-22 mandates GSA repository for AI procurement data
• Four agencies agreed to GAO’s policy update recommendations
• Failure to share insights risks costly repeat procurement mistakes

Pulse Analysis

Federal AI procurement is accelerating, yet the mechanisms for learning from past purchases remain fragmented. The GAO’s recent review highlights that while agencies such as the Department of Defense and the Department of Veterans Affairs are expanding AI deployments, they lack a formal process to capture what works and what doesn’t. This gap is especially stark given the surge in AI use cases—more than a 100% increase from 2023 to 2024—creating a pressing need for institutional memory to avoid duplicated effort and hidden costs.

The core of the problem lies in policy inertia. Current acquisition guidelines do not require agencies to systematically document lessons learned, leaving officials without the data needed to evaluate proposals, estimate total ownership costs, or protect sensitive government data. An April 2025 Office of Management and Budget memorandum (M-25-22) seeks to remedy this by directing agencies to contribute to a GSA‑run AI acquisition repository. GAO’s four recommendations focus on embedding collection and sharing mandates into agency policies, a step that all four surveyed departments have endorsed but have yet to operationalize.

Implementing a unified repository promises tangible benefits: faster procurement cycles, clearer cost structures, and reduced risk of repeat mistakes that could waste taxpayer dollars. Moreover, a shared knowledge base can foster cross‑agency collaboration, enabling smaller departments to leverage best practices from larger, more experienced counterparts. As the federal government strives to harness AI responsibly, establishing robust feedback loops will be essential for scaling innovation while maintaining fiscal discipline and security standards.