A leadership brief on the adoption, trust, and modernization forces pushing public-sector institutions to govern AI now — validated against workforce, budget, and program data for government executives.
Seven in ten public servants are already using AI at their desks — often without a policy telling them how. That gap between grassroots adoption and institutional readiness is the defining risk facing government leaders today: not whether staff will use AI, but whether the agency governs how they do.
The evidence shows a public sector accelerating fast on enthusiasm and slower on the enablement that turns adoption into results citizens can see. This report examines the signals driving that gap — and the operating discipline that closes it before budgets tighten and trust erodes.
Built to forward — use the rail on the left to send this to your CIO, department head, or oversight board and bring the case for backing your AI initiative.
Over 70% of public servants worldwide now use AI in their work, yet only 18% say their government is using it effectively.[1] That 52-point gap is the single most important fact for public-sector leadership: staff have already decided AI is useful, and the institution has not yet decided how to govern that use.
In countries with clear guidance and leadership backing, 91% of public servants feel confident using AI. Where rules and support are unclear, confidence collapses.
— Public Sector AI Adoption Index, 3,335 public servants across 10 countries, 2026
Even the UK — with a national AI strategy, a Digital Centre of Government, and a published AI Playbook — ranks sixth of ten countries surveyed, scoring 47 of 100.[2] Strategy documents are not the bottleneck; enablement on the ground is.
The global AI-in-government market is projected to grow from roughly $22.4 billion in 2024 to more than $98 billion by 2033 — a 17.8% CAGR.[3] In the U.S., federal agencies have committed more than $3 billion in civilian AI spending in the latest budget cycle alone, alongside a separate $13.4 billion Defense Department request for AI and autonomous systems in fiscal 2026.[4]
Efficiency mandate. Every surveyed federal agency leader is running efficiency initiatives in FY2026, with 43% citing AI and machine learning investment as a primary lever.[5]
Persistent barriers. Budget constraints (34%), outdated technology infrastructure (32%), and lack of skilled personnel (31%) remain the top hurdles agencies report — none of which a model upgrade alone solves.[5]
Legitimacy risk. A recent GAO review of how DOD, DHS, GSA, and the VA acquire AI capability found adoption metrics that look strong on paper but mask uneven acquisition discipline underneath.[4]
Trust exposure. Unlike private-sector AI, public-sector deployments answer to citizens and oversight bodies directly — governance gaps become legitimacy problems, not just operational ones.
In a survey of 2,000 US public-sector workers, 37% describe their agency's AI integration as advanced — embedded in multiple mission-critical processes — with another 32% actively developing it; only 6% report no integration at all.[6] 80% say AI feels empowering in their daily role.[1]
Fourteen percent of agencies are re-evaluating AI investments over budget or staffing concerns — and 12% specifically cite a lack of measurable ROI.
— Appian survey of 2,000 US public-sector workers, 2026
Constituent services and case processing. The highest-confidence starting point — cutting the time between a citizen's request and a resolved case.
Document and records processing. High-volume, well-bounded work with clear before/after time savings and modest doctrinal risk.
Fraud, waste, and improper-payment detection. Directly defensible to oversight bodies and taxpayers, with measurable dollar recovery.
Cybersecurity. Cited by 44% of federal leaders as a top efficiency priority for FY2026 — often the first place trust and modernization budgets meet.[5]
The index's own finding is the clearest guidance available: enthusiasm, empowerment, enablement, embedding, and education are mutually reinforcing — enthusiasm without enablement stalls adoption, and tools without training create risk rather than results.[1] The agencies scoring well share published, leadership-backed guidance; approved tools staff can point to; and training that closes the confidence gap the index measures.
Grassroots adoption, efficiency mandates, and modernization budgets are converging at once — but so is the risk of governing none of it well. Agencies that publish clear guidance, name accountable owners, and train staff now will convert today's enthusiasm into the effectiveness citizens and oversight bodies expect.
The institutions that close this gap first won't just modernize service delivery — they'll set the standard the next decade of public-sector AI is measured against.
Download the PDF, forward it to your leadership team, and bring your agency's AI initiative to the Executive Accelerator.