OMB Issues Revised Policies on AI Use and Procurement by Federal Agencies

The law firm Hunton reports that on April 3, 2025, the OMB, a White House office, put out two revised policies governing how federal agencies use and buy artificial intelligence. They are memos M-25-21 and M-25-22, which support Executive Order 14179 of January 23, 2025. They largely replace Biden-era memos, including M-24-10 from March 28, 2024.

Compared with the earlier approach, the new memos take a pro-innovation line that favors faster adoption and less bureaucracy. They give agency leaders more responsibility for governance, require public transparency, allow waivers for high-impact uses and transparency requirements when justified, and strongly prefer American-made AI and talent. M-25-21 covers AI developed, used or acquired by or for agencies, but not AI inside National Security Systems.

Under M-25-21, agencies must appoint Chief AI Officers within 60 days, and CFO Act agencies must set up governance boards within 90 days and publish strategies for removing barriers within 180 days. High-impact AI, meaning output that is a principal basis for decisions with legal, material, binding or significant effects on rights or safety, requires pre-deployment testing, impact assessments, human oversight, remedies or appeals, and pausing noncompliant use. Agencies must publish AI use case inventories at least annually and report waivers with justifications.

M-25-22 addresses procurement. It encourages buying U.S.-developed AI, requires compliance with privacy and intellectual property rules, and calls for contract terms barring vendors from using government data to train or develop models without consent. Contracts should guard against vendor lock-in, allow regular monitoring of performance and risk, and meet the high-impact minimum practices. Within 200 days GSA and OMB will build an online repository of procurement resources.