A new report from the Brookings Institution reveals that while the U.S. federal government has accelerated its adoption of artificial intelligence, significant hurdles remain in workforce capacity and risk culture. Despite bipartisan priority placed on AI integration over the past decade to enhance national security and service delivery, progress is unevenly distributed across agencies. Research analyzing data from 2023 to 2025 shows a marked increase in AI usage. In 2023, only 21 agencies, all large or midsize, reported using AI. By 2024, the number rose to 41 agencies, including 11 small agencies. In 2025, these 41 agencies documented over 3,600 distinct AI projects, representing a 69% increase from the previous year and five times the volume seen in 2023. While applications range from back-office automation to critical mission areas like health services and law enforcement, adoption remains highly concentrated. In 2025, five large agencies accounted for more than half of all federal AI use, with large entities employing over 15,000 staff driving three-quarters of total adoption. Valerie Wirtschafter, the Brookings fellow who authored the report, notes that despite this growth, AI-focused workers remain a tiny fraction of the federal technological workforce. Several structural bottlenecks are slowing broader implementation. The federal hiring process is often slow, and career advancement pathways for technologists are limited. This challenge was exacerbated following the second Trump administration, which laid off nearly 300,000 federal workers. Wirtschafter suggests these layoffs likely removed promising AI talent who were still on probation, causing a significant drop in AI-related job listings and undermining recruitment efforts. Beyond staffing issues, a deeply entrenched risk-averse culture discourages experimentation. The opacity of many AI algorithms, often described as black boxes, further erodes trust, particularly for sensitive government operations. Additionally, the increasing politicization of large language models presents a unique barrier. For instance, tools like Grok have been criticized for reflecting the political values of their creators, while others, such as Anthropic's Claude, have faced procurement hurdles due to perceived supply chain risks. Outdated infrastructure and legacy equipment also continue to hinder technological integration across the sector. To address these challenges, the report recommends specific reforms aimed at modernizing hiring practices, improving workforce retention, and fostering a culture that balances security with innovation. Wirtschafter emphasizes that while the federal government has made strides, a long path remains to achieve widespread, effective AI integration. The gap between large agencies that are scaling AI aggressively and smaller entities that are only beginning to experiment highlights the urgent need for a more inclusive and streamlined approach to federal technological adoption. Without resolving these cultural and structural deficits, the potential of AI to transform government efficiency and national competitiveness may remain unrealized.