Federal agencies are confronting a structural bottleneck in their AI expansion: the absence of standardized, interoperable data pipelines across departments limits the ability to move AI workloads from pilot to production at scale. New federal policy directives and a sharp rise in documented AI use cases are intensifying pressure on agencies to treat data infrastructure as a strategic priority.
Background
The Office of Management and Budget's latest AI inventory identified roughly 3,600 AI use cases across federal agencies, reflecting a nearly 70% year-over-year increase and underscoring how quickly AI is moving from experimentation to execution across government. Yet that expansion has exposed a foundational gap. As agencies shift from pilots to production, success will depend less on model access and more on whether they have the right data, governance, and operational discipline in place.
Many agencies operate with siloed systems, inconsistent standards, and weak data governance, undermining privacy, fairness, explainability, and accountability. According to NARA's 2023 FEREM report, only 61% of agencies were rated low-risk in their management of electronic records-indicating that a significant share still face gaps in data accessibility and governance that can prevent AI models from ingesting reliable inputs.
The policy landscape shifted substantially in April 2025. OMB Memorandum M-25-21 requires agencies to adopt enterprise AI strategies, designate Chief AI Officers, maintain public AI use-case inventories, and apply minimum risk management practices for "high-impact" AI while preserving privacy, civil rights, and security. A companion directive, M-25-22, governs procurement. It requires open and standard data formats and application programming interfaces (APIs) so that foundational components can support new use cases without proprietary dependencies-a direct response to vendor lock-in risks that have historically fragmented agency data ecosystems.
Details
Interoperability requirements are now embedded in federal AI acquisition policy. The OMB AI Use Memo advises covered agencies to "adopt procurement practices that encourage competition to sustain a robust Federal AI marketplace," including by "preferencing interoperable AI products and services." Contracts must clearly delineate ownership and intellectual property rights, data portability, and long-term interoperability. OMB's memorandum also requires agencies to include contract terms barring vendors from using non-public government data to train publicly or commercially available AI algorithms without explicit consent.
On the standards side, the migration of the National Information Exchange Model to the NIEMOpen framework under OASIS represents a structural step toward semantic interoperability. NIEMOpen, which originated as part of the federal response to information-sharing failures exposed on September 11, 2001, is now transitioning to open standards governance under OASIS. This transition signifies a key advancement in establishing open standards for data interoperability, reducing costs and enhancing the efficiency of information sharing across public sectors.1Federal Agencies Roll Out AI Strategy Plans: Takeaways for Government Contractors - Ogletree NIEMOpen-standardized data about individuals, organizations, locations, and activities from different agencies can feed into a shared knowledge graph; when a large language model operates in a GraphRAG architecture, it can query this graph to retrieve highly precise, factual context, significantly enhancing the accuracy and trustworthiness of AI-generated insights.2EXECUTIVE OFFICE OF THE PRESIDENT OFFICE OF MANAGEMENT AND BUDGET
Individual agencies are advancing on parallel tracks. HHS followed the OMB directives with its own AI strategy, built around a "OneHHS" model designed to consolidate data, infrastructure, and AI tools across the department. In September 2025, HHS rolled out ChatGPT to all its employees. The Department of State published an Enterprise Data and AI Strategy and reported approximately 45,000 active users on its secure generative AI platform, StateChat, as of September 2025. As AI tools move into production environments, the quality, governance, and stewardship of the underlying data become the primary line of defense against legal, ethical, and operational risk-a status codified by the 2025 AI Action Plan, which identifies high-quality data as a national strategic asset and elevates data professionals to central actors in responsible AI adoption.
Outlook
Covered agencies had until December 29, 2025, to revisit and update existing acquisition procedures to comply with OMB's requirements and ensure that their use of acquired AI conforms to M-25-21. Compliance plans from DHS, GSA, and the State Department signal that cross-agency alignment on data portability and metadata standards is now an operational requirement, not a design aspiration. As AI becomes more central to government missions, agencies that treat data as strategic infrastructure will be best positioned to scale trustworthy, mission-ready AI.
