Agencies Report over 3,000 AI Use Cases in 2025

Why It Matters

The inventory highlights the rapid expansion of AI in federal operations and signals shifting procurement dynamics as agencies rely heavily on commercial vendors while navigating security and policy constraints.

Key Takeaways

• 2025 inventory lists 3,611 AI use cases, up 105% from 2024.
• HHS reports 447 cases, the highest among all agencies.
• VA leads high‑impact AI with 215 of 445 such projects.
• Microsoft Copilot powers 102 federal AI deployments.
• Anthropic contracts face removal after Trump administration directive.

Pulse Analysis

The Office of Management and Budget's release of the 2025 Federal Agency Artificial Intelligence Use Case Inventory marks a watershed moment for government technology oversight. Compiled from 56 agencies, the catalog records 3,611 distinct AI applications—a 105 percent jump from the 1,757 cases reported in 2024. The surge reflects a cascade of executive orders and congressional mandates, including the 2023 National Defense Authorization Act and the February 2025 OMB memo that obliges agencies to disclose AI deployments. By centralizing this data, OMB creates a transparent baseline for future budgeting, risk assessment, and legislative scrutiny.

Health and Human Services tops the list with 447 use cases, underscoring the department’s aggressive push toward internal chatbots, predictive analytics, and clinical data summarization. Close behind are NASA, the Veterans Affairs (VA), Energy, and Justice departments, each reporting between 300 and 425 projects. Of the 445 high‑impact applications, the VA accounts for 215, ranging from voice‑bot call‑center modernization to AI‑driven anomaly detection. These high‑impact initiatives illustrate how AI is moving beyond experimental pilots to core service delivery, especially in health, veterans’ care, and scientific research.

Commercial vendors dominate the federal AI supply chain, with Microsoft’s Copilot suite appearing in 102 reported deployments, cementing the tech giant’s role as the de‑facto provider for government‑grade generative tools. Anthropic’s Claude, cited in 25 cases, now faces a political reversal: a Trump administration directive ordered agencies to terminate contracts after the company refused to support surveillance or autonomous weapon applications. This tug‑of‑war between innovation and national‑security concerns signals that future procurement will be tightly coupled to policy compliance, prompting agencies to diversify vendors and strengthen governance frameworks.