Government AI Files

About

In 2020, Executive Order 13960 (later codified in the Advancing American AI Act) established a requirement for federal agencies to publish an annual inventory of their artificial intelligence use cases. However, this data is often difficult to access or search through, especially across different agencies. This website intends to make these files more accessible to the general public.

Sources

The initial data gets published by each individual agency on their own website. The Office of the Federal Chief Information Officer then consolidates the information into a single file. We perform our own merging of this data in order to standardize responses across years and across agencies.

About us

Contact us by email: gov-ai-files (at) seas.upenn.edu

Sorelle Friedler is the Shibulal Family Professor of Computer Science at Haverford College, a Nonresident Senior Fellow at The Brookings Institution, and the Chair of ACM's U.S. Technology Policy Committee. She previously served as Assistant Director for Data and Democracy in the White House Office of Science and Technology Policy where she co-authored the AI Bill of Rights and contributed to federal policy on AI. Her research focuses on how AI impacts society, and she is a Co-Founder of the ACM Conference on Fairness, Accountability, and Transparency (FAccT).

Danaé Metaxa is Raj and Neera Singh Term Assistant Professor at the University of Pennsylvania with joint appointments in Computer and Information Science and the Annenberg School for Communication, where they co-founded the Penn HCI group. Their research focuses on bias and representation in algorithmic systems across high-stakes domains including politics, employment, and advertising, with particular attention to impacts on marginalized communities.

Emma Lurie is a postdoctoral researcher at the University of Pennsylvania and incoming assistant professor at Barnard College with a J.D. from Stanford Law School and Ph.D. in Information Science from UC Berkeley, specializing in public interest technology at the intersections of platforms, democracy, and law. She conducts audits of AI systems and combines legal analysis with algorithm auditing to address how technology platforms can better serve democratic societies while protecting civil liberties.

Emma Fauser is an undergraduate student at Haverford College pursuing computer science and urban studies. Her interests broadly lie in the applied use of computing for social problems. She is also a founding employee at Modrinth, a startup supporting Minecraft content creators.

Qing He is a graduate student at the University of Pennsylvania with a background in Computer Science, Finance, and Architecture. Her research sits at the intersection of human-AI interaction and cognitive augmentation, with a focus on designing tools that support complex decision-making and knowledge work.

Identified Discrepancies

This section intends to list all of the known discrepancies between our merged data (as used on this website) and the official OMB consolidated file for 2025. We hope that these discrepancies can be resolved in a later revision, either published by the individual agencies or by the OMB.

  • The OMB file has five (5) extra entries that were not present in the individual agency files, and thus are not present in the search interface. For posterity, these extra entries are hosted here.
    • Department of Homeland Security: two (2) extra entries: DHS-9 and DHS-2544. Both are classified as retired and not high impact.
    • National Credit Union Administration: three (3) extra entries: NCUA-12-2025, NCUA-13-2025, and NCUA-14-2025. All three are classified as deployed and high impact.
  • The OMB file is missing at least twenty-five (25) entries:
  • The OMB file has twenty-seven (27) entries that were supposed to be withheld, either due to a disclosure risk or because disclosure is prohibited by law. Eighteen (18) of these were not present in the individual agency files, and thus are not present in the search interface. For posterity, these withheld entries are hosted here.
    • National Aeronautics and Space Administration: fifteen (15) entries that were supposed to be withheld.
    • Environmental Protection Agency: three (3) entries that were supposed to be withheld due to a disclosure risk were included in the OMB file.
    • Nine (9) other entries that are present in the search interface from the Department of Labor, Federal Energy Regulatory Commission, and National Archives and Records Administration.
  • The OMB file has entry DHS-2552 under "Pre-deployment" status instead of "Deployed", and is also missing several of the agency data fields.

If you find more discrepancies, please notify us using the contact email above.