▶ Article reader Listen + reading controls ⌄

LISTEN + READ YOUR WAY

Article reader

Preparing the reader…

▶ Play article ← → Stop

0:00 0:00

Speed 0.75× 1× 1.25× 1.5× 1.75×

Reading settings

Typeface Field Notes Atkinson Hyperlegible Lexend OpenDyslexic

Text size

A−

100%

A+

Reading surface Field Notes dark Paper light Warm sepia High contrast Text spacing Standard Comfortable

Reset reading preferences

The federal AI inventory is becoming an operating instrument

The United States Government Accountability Office's (GAO) July 29 report finds that generative artificial intelligence (AI) use cases at the federal agencies under review have increased nearly ninefold between 2023 and 2024, from 32 to 282. Across the same inventories, reported AI use cases of all kinds have nearly doubled.

That is more than a growth statistic. It is a warning about the management problem arriving behind the technology.

An inventory sounds administrative: a list collected for oversight and filed once a year. But an accurate inventory can become one of the most useful instruments in a public-sector AI operating model. It can show where demand is growing, where different teams are solving the same problem, and where risks or infrastructure gaps are accumulating.

The question is whether the inventory describes living systems or dead entries.

Counted is not the same as understood

A useful entry needs more than a project name and a short description. It should identify the decision or task being supported, the people affected, the data and model dependencies, the operating environment, the accountable owner, and the evidence required for continued use.

It should also record status honestly. A proposal, experiment, limited pilot, operational capability, and retired system are not interchangeable. Treating them as one category inflates apparent adoption while obscuring where value actually reaches users.

GAO finds that agencies are navigating policy, technical-resource, budget, and appropriate-use challenges while the technology continues to change. Those constraints are not annotations around the inventory. They are part of what the inventory should make visible.

The inventory should connect the organization

The Office of Management and Budget (OMB) requires agencies to inventory AI uses and centrally track high-impact systems. The greater opportunity is to use that common record to connect professions that otherwise meet too late:

• mission owners can see which operational problem the use case addresses;
• data teams can trace sources, rights, and quality;
• cybersecurity teams can identify exposure and control needs;
• acquisition professionals can see vendor and contract dependencies;
• evaluators can locate test results and monitoring plans;
• and workforce leaders can anticipate changes to roles and training.

This is a knowledge-management function. The inventory becomes a shared map of commitments, not merely a count of tools.

Build it for decisions

A practical inventory should answer four recurring questions:

1. Where should the organization invest next?
2. Which capabilities can be reused instead of rebuilt?
3. Which systems need additional assurance or executive attention?
4. What has the organization learned from systems it stopped?

That requires routine updates, clear data ownership, controlled vocabularies, and links to authoritative artifacts. It also requires an expiration mechanism. If an entry has not been validated by its owner within a defined period, its uncertainty should be visible.

The National Institute of Standards and Technology AI Risk Management Framework emphasizes context, measurement, management, and governance. An inventory can connect those activities across a portfolio, but only if teams use it during prioritization and review.

Federal AI adoption is already too large to manage through memory and personal networks alone. A living inventory will not solve every governance problem. It can, however, give leaders a reliable place to see what exists, who is responsible, what evidence is missing, and where one program's lesson could save another program months.

Sources and research trail

• United States Government Accountability Office, Generative AI Use and Management at Federal Agencies (July 29, 2025).
• Office of Management and Budget, M-25-21: Accelerating Federal Use of AI through Innovation, Governance, and Public Trust (April 3, 2025).
• National Institute of Standards and Technology, Artificial Intelligence Risk Management Framework 1.0 (2023).
• Alavi and Leidner, “Review: Knowledge Management and Knowledge Management Systems” (2001).