Federal Government Slows AI Adoption Despite Three-Year Push
The U.S. government is using artificial intelligence more than it did three years ago, but progress remains uneven and constrained by hiring gaps, risk-averse decision-making, and doubts about AI's reliability. New research from the Brookings Institution found that while federal AI adoption has accelerated, five agencies account for more than half of all use.
Federal agencies reported 3,600 distinct AI projects in 2025, up 69 percent from 2024 and five times the number from 2023, according to analysis of government data and interviews with technology specialists across eight agencies. The work ranges from automating routine tasks to delivering benefits, providing medical services, and supporting law enforcement.
But the gains mask significant disparities. Large agencies with more than 15,000 employees accounted for over three-quarters of AI use in 2025. Small and midsize agencies are experimenting with AI, but large agencies are scaling faster.
Hiring Remains the Primary Bottleneck
Federal agencies struggle to recruit and retain AI specialists. The government's slow hiring process and limited advancement opportunities for technologists make it difficult to build in-house capacity.
The recent wave of federal workforce reductions has made the problem worse. The second Trump administration laid off nearly 300,000 federal workers last year. Many recent AI hires were still probationary and likely lost their jobs, undermining recruitment efforts for future AI talent.
Culture and Trust Issues Slow Wider Adoption
Beyond hiring, federal agencies operate within a risk-averse culture that discourages experimentation. Agencies handling sensitive health or security data face additional barriers.
The opacity of AI systems creates another obstacle. When agencies cannot explain how an AI system reached a conclusion, they hesitate to deploy it for sensitive work. This is particularly true for large language models, which have become politicized.
Grok, developed by Elon Musk's xAI, has a documented history of reflecting political values and generating problematic content. The Department of Defense has labeled Anthropic's Claude a "supply chain risk" following contract disagreements with the agency. These disputes underscore how political concerns can slow AI adoption across government.
What Agencies Need to Move Faster
Research identified several steps to accelerate adoption:
- Streamline hiring for AI-focused positions
- Create career advancement pathways for technologists
- Treat AI literacy as a core job requirement and invest in training
- Share success stories across agencies to build confidence
- Increase transparency about how agencies use AI
- Focus investment on high-impact projects that directly improve service delivery
The Biden and Trump administrations both prioritized federal AI adoption to improve service delivery, support data-driven decision-making, and strengthen national security. The gap between policy goals and actual progress shows that intent alone does not drive organizational change.
For government employees working on technology projects, the current environment presents both constraints and opportunities. Hiring limitations mean competition for positions is less fierce. Risk-averse cultures create space for careful, well-documented pilots that demonstrate value before full rollout.
Learn more about AI for Government adoption strategies and the role of Generative AI and LLM systems in federal workflows.
Your membership also unlocks:
