The federal government’s annual accounting of artificial intelligence use cases reflects a steady acceleration in adoption. Agencies across Washington have cataloged more than 2,500 AI applications, a sharp increase from prior inventories, as AI shifts from limited pilot programs into operational systems embedded in day-to-day government work.
The most recent consolidated federal inventory publicly reported in 2024 documented total of 2,133 publicly reportable use cases across 41 federal agencies, according to FedScoop on the Office of Management and Budget’s annual tally of AI deployments. Updated totals expected this year are projected to climb again as more use cases move from pilot to production and additional agencies expand their disclosures.
Federal Chief Information Officer Greg Barbaccia has emphasized that the annual inventory is designed not only to track adoption but also to standardize governance and oversight as AI deployments expand across agencies.
Nearly 90% of federal government IT leaders and influencers surveyed said their agencies are planning to use or are already using AI, according to Google Public Sector research based on a survey of 250 federal IT professionals across civilian and defense agencies. The findings suggest that AI adoption in government is broader than conventional narratives often assume.
According to that research, the most common deployments include document and data processing, workflow and process automation, and decision support systems. Agencies also report using AI to analyze large datasets, monitor network activity in real time, streamline case management, and support fraud detection efforts. In many cases, AI is being applied where data volumes are large and manual review is resource intensive.
Rewriting the Government IT Playbook
At the State Department, AI deployment has required a fundamental shift in how technology is introduced to the workforce. Kelly Fletcher, the department’s chief information officer, described moving away from the traditional federal IT rollout model.
“We are actually doing something a little bit different with AI than what we’ve done with IT historically,” Fletcher said. “Historically, we would roll out technology, we would test it really carefully, and then it would be sort of like etched in stone.”
Instead of waiting for a fully polished release, the State Department launched its generative AI chatbot, State Chat, to an initial group of 300 users more than a year and a half ago. The early version was intentionally iterative.
That pilot group expanded to roughly 3,000 users who provided continuous feedback. Fletcher said prompt changes and feature updates were often noticed by users before analytics reflected the shift. The tool was eventually rolled out more broadly across the department’s roughly 100,000 employees.
“These users both affect what are the next capabilities that we add to the chatbot, but then also identify problems,” Fletcher said, describing a feedback-driven model that treats AI as an evolving capability rather than a static product.
Governance and Workforce Catch-Up
The U.S. Army is confronting a related but distinct issue: workforce readiness. Kris Saling, CTO for the Office of the Assistant Secretary of the Army for Manpower and Reserve Affairs, said the Army’s earlier AI strategies predate the rapid rise of generative AI tools.
“Strategy was written back in 2018 for the most part for the Army,” Saling said. “Since that time, November 2022 happened. We have ChatGPT. We have rapid proliferation of generative AI.”
The pace of change has altered assumptions about how many developers and end users are building and deploying AI-enabled tools internally.
“We figured out how to rapidly proliferate AI,” Saling said. “What we haven’t figured out how to do is how to rapidly train people how to use it responsibly.”
Saling emphasized that the challenge is not simply technical training but ensuring personnel understand the domain context in which AI systems operate. The Army is working to align previously siloed systems and strategies into a more unified approach.
The expansion of federal AI use cases does not mean adoption is frictionless. In Google Public Sector’s survey, security and adversarial risks were cited by 48% of respondents as the top barrier to scaling AI. Reliability concerns followed at 35%, while workforce-related concerns, including potential disruption and skills gaps, were also flagged by agencies navigating deployment at scale.
