The Office of Management and Budget (OMB) has issued two new memoranda—M-25-21 and M-25-22—designed to guide the responsible adoption and acquisition of artificial intelligence (AI) across federal agencies.
Together, these memoranda aim to modernize federal operations through AI while promoting responsible governance and accountability. Robert Shea, former Associate Director of OMB, noted the memos introduce familiar but more actionable requirements, such as designating AI leadership roles and implementing structured risk management processes.
M-25-21: Framework for AI Use in Government
Memorandum M-25-21 outlines a government-wide framework aimed at accelerating AI adoption while maintaining safeguards related to privacy, civil rights, and civil liberties. It mandates that agencies designate a Chief AI Officer (CAIO), develop and publish AI strategies, assess their AI maturity, and share AI models and data to reduce duplication.
The memo introduces a category for “high-impact AI” use cases—those with potentially significant effects on rights or safety—which must undergo risk assessments, testing, and ongoing monitoring.
Agencies must also maintain transparency through regular reporting, including inventories of AI use and risk compliance updates.
Read the full memorandum here. OMB has also provided a fact sheet related to this memorandum. Read it here.
M-25-22: AI Acquisition Guidance
Memorandum M-25-22 complements this framework by providing guidance on AI acquisition. It directs agencies to adopt open standards to avoid vendor lock-in, use performance-based contracting approaches, and integrate protections for privacy and intellectual property.
The memo encourages agencies to engage cross-functional teams when acquiring AI systems and to ensure procurement practices support interoperability and efficient oversight. It also highlights the importance of favoring AI tools developed in the United States when consistent with law and policy.
Read the full memorandum here.