The Interagency AI Working Group publishes a comprehensive Federal AI Use Case Taxonomy covering 287 documented AI applications across federal agencies with risk classification for each category.
The Interagency AI Working Group, co-chaired by NIST and the White House OSTP, published the Federal AI Use Case Taxonomy documenting 287 AI applications deployed or in active development across federal agencies. The taxonomy provides risk classifications, governance recommendations, and training requirements for each of the 14 high-level use case categories.
The 14 categories range from lower-risk administrative automation applications through higher-risk mission applications such as benefits eligibility determination, suspect identification, and threat intelligence analysis. Each category includes risk factors, required oversight mechanisms, and minimum training standards.
Benefits determination AI systems — used by SSA, VA, and HHS — represent the largest single use case category with 67 documented deployments. These systems receive the highest risk classification, requiring human-in-the-loop review for all adverse decisions and mandatory bias auditing at six-month intervals.
The taxonomy documents significant inconsistency in AI governance practices across similar use cases at different agencies, and recommended that agencies benchmark their governance practices against peers deploying similar use case categories.
GovAcademy's AI Governance for Public Sector Leaders course (GA-002) incorporates the Federal AI Use Case Taxonomy as a reference framework, and the AI Risk Register for Agencies course (GA-040) uses the taxonomy's risk classification methodology as the basis for its risk identification exercises.
The AI-Assisted Case Management course (GA-021) is particularly relevant for agencies implementing AI in the benefits determination and case management categories, covering the human-in-the-loop design requirements and bias mitigation practices that the taxonomy requires.
NIST has published the taxonomy as a companion document to the NIST AI RMF, providing federal practitioners with the sector-specific guidance that the framework itself does not provide.