What makes AI work in government? GGF study identifies five foundations for success

From detecting fraud and cutting time spent on administrative tasks to accelerating nuclear energy deployment and even monitoring crocodiles, governments are implementing AI in day-to-day operations. A new Global Government Forum study reveals what they are learning about making it work

Governments around the world are moving beyond AI experimentation to deployment – but turning promising pilots into lasting improvements to public services requires much more than getting the technology right.

New Global Government Forum research – titled AI that delivers: Practical foundations for government success – examines what governments need to do in order to turn AI’s potential into tangible results. It also highlights growing ambition to move beyond using AI to improve existing processes towards more transformative approaches that rethink how governments operate and deliver services.

The research draws on interviews with senior public servants from Australia, Spain, the United Kingdom and the United States, and examples featured span administrative support, compliance and fraud detection, health and citizen services, scientific and environmental applications, and the deployment of critical infrastructure.

Download the study: AI that delivers: Practical foundations for government success

Government AI in action

The applications spotlighted in the study demonstrate the breadth of ways governments are already using AI.

At the US Department of Health and Human Services, the Office of Inspector General is using AI to help prevent fraud, waste and abuse across the programmes it oversees, including Medicare and Medicaid. The US Department of Energy’s Office of Nuclear Energy has established an initiative to use AI to accelerate the rollout of nuclear energy technologies, as it works towards a target of quadrupling nuclear energy capacity to 400 gigawatts by 2050.

Elsewhere, the UK’s tax, payments and customs authority HMRC is using AI in complaints triage and routing, compliance casework and fraud detection. It says that AI and advanced analytics have already enabled the protection and recovery of £10bn (US$13.5bn) in tax between 2025 and 2026.

Applications in Spain designed to be reuseable across government include legislation analysis, machine translation and citizen-facing virtual assistants.

Meanwhile, Parks Australia, which is part of the Department of Climate Change, Energy, the Environment and Water, is combining drones, sensors and AI-enabled image analysis to support environmental management. Applications range from the Ghost Net and Marine Debris Initiative, where First Nations ranger groups use drones and AI to detect and prioritise the clean-up of marine debris across remote coastlines, to the use of AI to support crocodile and feral animal monitoring.

Join the webinar on 22 September to discuss the study’s findings

Five foundations for government AI

Sarah Wray, research editor at Global Government Forum, said: “Across the diversity of the applications, common lessons emerged. The report identifies five foundations for successful government AI: lead with the problem; use governance and trust as enablers; strengthen data foundations; build workforce capability; and design for scale.”

The study also highlights the steps that governments are taking to bolster these foundations, from using impact assessments to prioritise AI ideas to building secure environments for sensitive data and involving frontline staff in developing and testing tools. Other examples include establishing multidisciplinary teams, rolling out workforce-wide AI training and appointing dedicated AI leaders. The report also explores how governments are building continuous oversight into AI initiatives and setting clear criteria for deciding when pilots are ready to scale.

Preparing for the future and agentic AI

Looking to the future, the report finds that there is emerging interest in agentic AI and AI-supported workflows, including applications that could operate across organisational boundaries. This underpins the growing importance of strong foundations for deploying AI responsibly and effectively at scale.

SAS and Microsoft were Global Government Forum’s knowledge partners for the study. Jennifer Robinson, global public sector strategic advisor at SAS, said: “With a clear vision and strategy that emphasises data, governance, and workforce capability, you can guide your teams toward successful AI implementation. Establishing this framework will help your organisation identify where and how AI is best used and create the conditions for trusted outcomes.

“If done right, your organisation’s AI will facilitate the work of your staff and improve the lives of the people you serve – today and in the future.”

Interviewees for the study were:

• Dr Rian Bahran, deputy assistant secretary for nuclear reactors, Office of Nuclear Energy, US Department of Energy, United States
• Leandro Higueruela Jaroso, deputy assistant director, Artificial Intelligence Governance and Planning Division, State Secretariat for Digitalisation and Artificial Intelligence, Ministry for Digital Transformation and the Civil Service, Spain
• Andrew Jansen, director of AI and digital delivery, Parks Australia, Department of Climate Change, Energy, the Environment and Water, Australia
• Robert Owens, chief financial officer, Office of Inspector General, Department of Health and Human Services, United States
• Lucy Poole, deputy chief executive officer, Strategy, Planning and Performance Division, Digital Transformation Agency, Australia
• Giuseppe Sollazzo, deputy director, Chief AI Office, His Majesty’s Revenue and Customs, United Kingdom
• Keith Tayler, supervising scientist, Office of the Supervising Scientist, Department of Climate Change, Energy, the Environment and Water, Australia