Federal agencies are expanding their use of artificial intelligence, but a new Government Accountability Office report identified key challenges in buying and deploying the technology.
Leaders from GAO, the Pentagon and other agencies will explore the challenges in deploying AI in federal and mission-critical systems at the Potomac Officers Club's 2026 Digital Transformation Summit on April 22. The event is dedicating multiple panels to AI, reflecting the technology's growing adoption in the public sector to support operations and decision-making. You can still get your tickets here.
What Challenges Do Agencies Face in AI Procurement?
Officials interviewed for the report, issued Monday, cited several barriers to effective AI procurement, including limited access to technical experts such as data scientists who can evaluate vendor proposals and assess the performance of AI systems. Agencies also reported difficulty understanding AI-related costs.
GAO also found that agencies are not systematically collecting or sharing insights from past procurements. In 2025, the Office of Management and Budget issued guidance for the responsible procurement and use of AI, requiring agencies to submit best practices, knowledge and resources about AI acquisitions to a shared web-based repository. However, the General Services Administration, the Department of War, the Department of Homeland Security and the Department of Veterans Affairs cannot comply with the rule due to existing internal policies that prohibit them from collecting lessons learned.
GAO warned that agencies may miss opportunities to identify best practices or avoid repeating mistakes.
The watchdog urged agencies to update policies to require the collection and sharing of acquisition insights.
How Are Agencies Acquiring AI Technologies?
The report also detailed the various approaches agencies use to acquire AI capabilities. Some agencies pursue agency-directed procurements by issuing contracts for specific requirements, while others adopt vendor-driven approaches in which companies introduce AI capabilities without pre-defined needs.
Agencies also vary in how they structure agreements. While traditional contracts remain common, some organizations use alternative agreements outside the Federal Acquisition Regulation to acquire advanced capabilities. In addition, some agencies are opting for AI-as-a-service, in which the vendor provides capabilities on an ongoing basis, rather than acquiring software.
