AI Is Entering Government Faster Than Policies Can Keep Up

2 October 2026

Artificial intelligence is already making its way into government work, with public institutions using it to search for information, summarise documents, detect fraud, answer questions and support service delivery. World Bank Blogs

Uganda is among the countries highlighted in new World Bank research on the use of artificial intelligence in government. The study reports that the country has developed an AI-HR Assistant, a chatbot designed to respond to human-resource queries from within an electronic document management system. Elsewhere, governments are using AI to identify possible tax fraud, analyse medical images, support examination marking and help citizens navigate legal information.

As that adoption gathers pace, the policies and systems needed to guide the use of AI are not always developing at the same rate.

That gap is one of the main findings from a new World Bank study covering government use of artificial intelligence in 60 economies. The research gathered information directly from central digital agencies, ministries and officials responsible for government information systems.

AI use was already widespread among the governments surveyed, but much of it had not yet become part of formal institutional processes.

Forty-four per cent of internal government AI use involved individual public servants using the technology for tasks such as searching for information, summarising documents and obtaining basic assistance. The World Bank describes this as relatively informal use because the tools are being used by individuals rather than being built into established government workflows.

That distinction becomes important once AI begins handling government information or contributing to decisions. A public servant using an AI tool to shorten a report is one thing. Using the same technology to process administrative records, respond to citizens or support decisions affecting public services brings questions about privacy, accuracy, security and responsibility.

The survey shows that many governments are still working out those questions. Nearly two-thirds of the governments surveyed provide public servants with licences or official access to generative AI tools. Only 39 per cent had formal ministry-wide guidelines governing their use, while another 38 per cent said such guidelines were still being developed.

The World Bank refers to one consequence as "shadow AI", where employees use artificial intelligence without enough institutional guidance or oversight. This can leave uncertainty over what information may be entered into an AI system, when a human should review its output and who is responsible when AI contributes to a government decision.

The challenges are not the same everywhere because among low-income economies in the survey, 58 per cent identified the absence of AI policies, guidelines, frameworks or standards among their three biggest barriers to adoption. Data quality and availability were also repeatedly identified as obstacles across countries at different income levels.

Uganda's human-resource assistant is one example. Burkina Faso has developed an AI assistant to help people navigate legal information, while the Democratic Republic of Congo is using an AI system to support the correction of state examinations. The Philippines is applying predictive AI in tax administration and Thailand is using the technology in the detection of tuberculosis and lung cancer from chest X-rays.

These examples shift the discussion from whether governments will use AI to how they will use it well.

The University's Department of Computer Engineering and Informatics lists artificial intelligence, data science, cybersecurity, computer forensics and cyber-physical systems among its areas of expertise. It also offers a Master of Science in Artificial Intelligence and identifies collaboration with government and other organisations as part of its research work.

Researchers can examine how public institutions handle sensitive data when using AI, how outputs are checked before they influence decisions, how systems can be protected from security threats and whether public servants understand both the capabilities and limitations of the tools they are using.

An AI application may work during a small pilot but behave differently once it is connected to larger databases, used by hundreds of staff or placed inside a public service that people depend on every day. Researchers can test accuracy, reliability, bias and system performance under those conditions.

The human side is just as important because public servants need enough understanding of AI to know when its output is useful, when it requires checking and when human judgement should take priority. The World Bank argues that governments need stronger skills, better data systems and organisational structures that allow experimentation to become institutional learning.

These are questions that sit across several fields rather than inside computer science alone. Cybersecurity researchers can examine how government data is protected. Data scientists can study the quality and suitability of the information used to train or operate AI systems. Researchers in public administration can look at accountability and institutional processes, while legal and ethical research can examine what safeguards are needed when automated systems begin influencing public decisions.

Busitema's existing work in artificial intelligence gives the University a foundation from which some of these questions can be explored. Its AI research group brings together people from different academic backgrounds, while the Department of Computer Engineering and Informatics already works in areas closely connected to secure and responsible digital systems.

As AI becomes part of everyday government work, it is opening new opportunities to improve how institutions manage information, support staff and deliver public services. Research can help show how these systems perform in practice and how they can be integrated responsibly and effectively into public institutions.

Uganda's appearance in the World Bank study shows that this transition is already underway. The bigger task now is ensuring that the systems surrounding AI develop as quickly as the technology itself.

Read the original World Bank article, How are governments using AI? New evidence from around the world