OMB Memorandum M-25-21: Accelerating Federal Use of AI through Innovation, Governance, and Public Trust

Overview

The OMB Memorandum M-25-21, officially titled “Accelerating Federal Use of AI through Innovation, Governance, and Public Trust,” was issued by the Office of Management and Budget (OMB) on April 3, 2025. This pivotal document sets forth new requirements and comprehensive guidance for federal agencies regarding the development, use, and procurement of Artificial Intelligence (AI) technologies. It emerged as a direct response to President Trump's Executive Order 14179, “Removing Barriers to American Leadership in Artificial Intelligence,” signed on January 23, 2025. The core objective of M-25-21 is to foster a forward-leaning and pro-innovation approach to AI within the U.S. federal government, aiming to accelerate the adoption and acquisition of AI systems by reducing bureaucratic hurdles and maximizing American competitiveness in the global AI landscape.

This memorandum significantly updates and, in many aspects, replaces previous AI guidance, notably OMB Memorandum M-24-10, “Advancing Governance, Innovation, and Risk Management for Agency Use of Artificial Intelligence,” which was issued on March 28, 2024. While drawing heavily from its predecessor, M-25-21 introduces a distinct shift in emphasis, prioritizing accelerated AI deployment and innovation. It directs agencies to enhance public services, bolster national security, and improve economic competitiveness through strategic AI integration, all while maintaining robust safeguards for civil rights, civil liberties, and privacy. The memorandum underscores the importance of leveraging American AI talent and tools, positioning the United States at the forefront of AI development and application.

Definitions

OMB Memorandum M-25-21 defines key terms to ensure consistent application across federal agencies. Central to its framework is the concept of “High-Impact AI” systems. These are generally understood as AI systems whose outputs serve as a principal basis for decisions or actions that have a legal, material, binding, or significant effect on an individual's rights or safety. This includes, but is not limited to, AI applications related to national security or protected speech. The memorandum maintains many of the risk management practices for what M-24-10 referred to as “Safety and Rights Impacting AI,” but M-25-21 streamlines the definition, narrowing the number of “presumed” high-impact use cases and consolidating specific areas under a more general “access to government benefits” section.

The term “agency” within the context of M-25-21 refers broadly to executive departments, government corporations, and independent regulatory agencies, as defined in 44 U.S.C. § 3502(1). However, the memorandum explicitly exempts AI systems that are part of a “National Security System,” noting that such uses are governed by separate policies and guidelines. This distinction is crucial for understanding the scope of the memorandum's applicability, allowing for specialized regulations to govern AI in critical national security contexts while providing a unified framework for civilian federal agency AI use. The guidance also implicitly defines “AI innovation” as the timely uptake and deployment of AI across federal agencies to maximize public benefit and improve government services.

Governance and Institutional Framework

The memorandum establishes a robust governance and institutional framework designed to facilitate the responsible and accelerated adoption of AI across federal agencies. A cornerstone of this framework is the requirement for each agency to appoint a Chief AI Officer (CAIO) within 60 days of the memorandum's issuance. These CAIOs are tasked with establishing comprehensive processes for managing “high-impact AI” systems, coordinating AI use within their respective agencies, and ensuring compliance with applicable laws and government-wide guidance. Their roles are critical in driving AI innovation while upholding safeguards for civil rights, civil liberties, and privacy.

Further strengthening AI governance, M-25-21 mandates the convening of an Agency AI Governance Board within 90 days for agencies subject to the Chief Financial Officers (CFO) Act. These boards are responsible for coordinating the use of AI within the executive branch, removing barriers to AI adoption, and managing associated risks. The memorandum also establishes a Chief AI Officer Council, intended to coordinate AI use across all covered agencies, promote shared templates and best practices, and facilitate interagency information sharing. This multi-layered governance structure aims to ensure a cohesive, effective, and accountable approach to AI implementation across the federal government.

Key Focus Areas

OMB Memorandum M-25-21 outlines three paramount key focus areas: innovation, governance, and public trust. The memorandum strongly advocates for a “forward-leaning and pro-innovation” approach, encouraging agencies to rapidly adopt and acquire AI technologies by reducing bureaucratic burdens and streamlining processes. This pro-innovation stance is geared towards maximizing U.S. competitiveness in AI and ensuring that the American public receives the maximum benefit from AI adoption, including improved public services and enhanced government efficiency. Agencies are directed to identify and remove barriers to the responsible use of AI, fostering an environment where AI solutions can be developed and deployed with agility.

Alongside innovation, robust governance is emphasized as essential for managing the risks associated with AI. The memorandum details requirements for AI governance efforts, risk management practices, and interagency coordination, particularly for “high-impact AI” systems. Public trust is the third critical pillar, underscored by mandates for transparency measures. Agencies are required to demonstrate AI risk mitigation, use, value, and efficiency to the public. Furthermore, the memorandum expresses a strong preference for American-made AI tools and services and places a significant focus on developing and retaining American AI talent, thereby reinforcing national leadership in AI innovation.

Implementation Framework

The implementation framework articulated in M-25-21 requires federal agencies to undertake several strategic initiatives to integrate AI responsibly and effectively. Agencies are directed to develop comprehensive AI strategies that identify and address barriers to the responsible use of AI and outline plans for achieving enterprise-wide improvements in AI application maturity. These strategies must include detailed plans for sharing AI functions, investing in AI technology, and upskilling existing staff to enhance internal AI capabilities. The memorandum encourages agencies to leverage and scale existing tools, processes, and resources for AI governance to avoid creating unnecessary bureaucracy.

A critical component of this framework involves the generation of AI compliance plans. Agencies subject to the CFO Act are mandated to publish these strategies within 180 days of the memorandum's issuance. These plans must detail how agencies will implement the directives of M-25-21, including specific actions for risk management, transparency, and innovation. The memorandum also emphasizes the importance of data governance and management, encouraging agencies to actively engage in the reuse of data and AI assets to save taxpayer dollars and foster government-wide collaboration. This holistic framework aims to ensure that AI adoption is not only accelerated but also strategic, secure, and aligned with public values.

Monitoring and Evaluation

OMB Memorandum M-25-21 establishes clear requirements for the monitoring and evaluation of AI systems deployed by federal agencies, particularly those categorized as “high-impact AI.” Agencies are mandated to conduct pre-deployment testing to identify both potential risks and benefits of each AI use case before its full implementation. This proactive approach ensures that potential issues are identified and addressed early in the development cycle.

Furthermore, the memorandum requires agencies to complete AI Impact Assessments both before and throughout the deployment of AI systems. These assessments are crucial for evaluating the intended purpose, expected benefits, performance of the AI model, and the ongoing impacts of its use on individuals and society. To ensure transparency and accountability, agencies must at least annually inventory and publicly publish their AI use cases. They are also required to publicly report risk determinations and any waivers granted from minimum practices for high-impact AI, along with a clear justification for such waivers. This comprehensive monitoring and evaluation framework aims to ensure continuous oversight and responsible AI deployment.

Penalties, Liability, and Appeals

While OMB Memorandum M-25-21 focuses primarily on establishing a framework for responsible AI use and governance, it also addresses the critical aspect of accountability and redress for individuals affected by AI-enabled decisions. The memorandum explicitly requires agencies to offer remedies or appeals for individuals impacted by decisions made or significantly influenced by AI systems. This provision is vital for upholding civil rights and civil liberties in the context of increasing AI adoption in government services.

Although the memorandum does not detail specific penalties for non-compliance by agencies, its binding nature on executive branch agencies implies that adherence to its directives is a mandatory aspect of federal operations. The emphasis on robust risk management practices, transparency, and human oversight serves as a preventative measure against potential harms. By mandating mechanisms for appeal and redress, M-25-21 aims to ensure that individuals have avenues to challenge and seek remedies for adverse outcomes resulting from federal AI use, thereby reinforcing public trust in government AI initiatives.

Relationship to Other Instruments

OMB Memorandum M-25-21 operates within a broader legal and policy landscape, building upon and superseding previous guidance while aligning with overarching executive and legislative mandates. It was issued as a direct implementation of President Trump's Executive Order 14179, “Removing Barriers to American Leadership in Artificial Intelligence,” signed on January 23, 2025. This Executive Order provided the foundational directive for accelerating AI adoption and innovation across the federal government.

The memorandum explicitly rescinds and replaces OMB Memorandum M-24-10, “Advancing Governance, Innovation, and Risk Management for Agency Use of Artificial Intelligence,” issued by the Biden Administration on March 28, 2024. While M-25-21 draws heavily from its predecessor, maintaining many core requirements for AI governance and risk management, it introduces a distinct “forward-leaning and pro-innovation” approach. Furthermore, M-25-21 is consistent with the AI in Government Act of 2020 and the Advancing American AI Act, reinforcing the legislative intent for responsible AI use within the federal sector. This layered approach ensures continuity in foundational AI principles while adapting to evolving policy priorities and technological advancements.

International Alignment

OMB Memorandum M-25-21, while primarily focused on domestic federal agency use of AI, implicitly positions the United States within the global AI landscape by emphasizing American leadership and competitiveness. The underlying Executive Order 14179, which M-25-21 implements, aims to advance the United States' global AI dominance. This focus suggests an awareness of international developments and a strategic intent to ensure the U.S. remains at the forefront of AI innovation and application.

The memorandum's approach, which encourages accelerated AI adoption and acquisition by reducing bureaucratic burdens, can be seen in contrast to some international regulatory frameworks, such as the EU AI Act. The EU AI Act, for instance, has a strong emphasis on risk-management practices and prohibitions on certain AI systems. While M-25-21 is legally binding only on U.S. federal agencies, its emphasis on promoting American AI tools and talent, alongside robust safeguards, contributes to the broader international discourse on responsible AI development and deployment, influencing global standards through leadership and example rather than direct harmonization.

Implementation Timeline

Sources and References