AI in Government: Public Sector Transformation in MENA

Government adoption of artificial intelligence across the Middle East and North Africa represents a distinctive model of technology deployment, one characterised by ambitious national strategies, substantial state investment, and centralised coordination that contrasts sharply with the more fragmented approaches typical of many Western democracies. The UAE government has articulated its intention to become the world’s leader in AI by 2031, backing this aspiration with a dedicated Minister of State for Artificial Intelligence—the first such cabinet position globally—and coordinated initiatives across federal and emirate-level authorities. Saudi Arabia’s Vision 2030 positions AI as central to economic diversification, with the National Strategy for Data and AI directing billions of dollars toward capability development. These national strategies share common themes: recognition that government must lead AI adoption to catalyse broader ecosystem development, belief that coordinated national approaches offer competitive advantages over market-led adoption, and acceptance that public sector transformation provides both direct benefits and demonstration effects that encourage private sector investment.

The scope of government AI applications under consideration or deployment across MENA spans virtually every domain of public administration. Immigration authorities have implemented AI-powered facial recognition and document verification systems at airports and border crossings, promising reduced wait times and enhanced security through biometric matching. Healthcare ministries have deployed AI diagnostic support systems in public hospitals, extending specialist capabilities to facilities lacking human expertise. Police agencies have adopted predictive analytics tools that attempt to anticipate crime patterns and allocate resources proactively. Municipal authorities use AI for traffic management, optimising signal timing to reduce congestion based on real-time flow analysis. Education ministries have piloted adaptive learning platforms in public schools, while labour ministries have experimented with AI matching systems that connect job seekers with opportunities. The breadth of these applications reflects both genuine potential for public service improvement and political imperative to demonstrate progress on strategic AI priorities.

Yet the gulf between announcement and implementation, between pilot and scale, often proves substantial in government AI initiatives worldwide, and MENA governments are not immune to this pattern. A McKinsey analysis of government AI readiness found that while several MENA governments rank highly on AI strategy and investment metrics, they score lower on implementation maturity and outcome measurement. The challenges are familiar to public sector technology initiatives everywhere: procurement processes ill-suited to rapidly evolving AI capabilities, legacy IT systems that resist integration with modern applications, workforce skills gaps that limit effective deployment and oversight, and accountability structures that may discourage the risk-taking that innovation requires. Understanding these implementation realities—rather than accepting strategic pronouncements at face value—is essential for realistic assessment of AI’s current and potential impact on MENA government operations and public services.

Smart Cities and the Infrastructure of Surveillance

Smart city initiatives represent the most visible and consequential domain of government AI deployment across MENA, combining genuine service improvements with surveillance capabilities that raise fundamental questions about privacy, autonomy, and the relationship between citizens and state. Dubai’s smart city programme, launched in 2014, has deployed AI across transportation, energy, public safety, and government services, creating an integrated urban management system that collects and analyses data from sensors, cameras, and citizen interactions across the city. Saudi Arabia’s NEOM, the $500 billion development rising from the desert in the Kingdom’s northwest, promises to be the world’s most technologically sophisticated city, with AI managing everything from autonomous transportation to building systems to the collection and treatment of waste. These mega-projects attract global attention and investment while raising questions about whether the surveillance infrastructure necessary to operate AI-managed cities is compatible with human flourishing and fundamental freedoms.

The surveillance dimension of smart city AI cannot be dismissed as merely hypothetical concern or Western projection. Facial recognition systems deployed in public spaces across Gulf cities can identify individuals moving through urban environments, creating records of presence and movement that governments can access for purposes ranging from legitimate security functions to more troubling applications. Amnesty International research has documented concerns about surveillance technology deployment in the region, while Carnegie Endowment analysis has tracked the spread of AI surveillance capabilities across MENA governments. The argument that citizens who “have nothing to hide” need not fear surveillance fails to reckon with the chilling effects that awareness of observation imposes—the way that knowledge of being watched constrains expression, association, and behaviour in ways that diminish human flourishing even absent any specific punitive action. For residents of MENA cities, the tradeoff between the convenience and efficiency benefits of smart city AI and the privacy costs of the surveillance infrastructure that enables it is rarely explicitly presented or genuinely chosen.

Responsible approaches to smart city AI must navigate the tension between service improvement and surveillance risk rather than pretending the tension does not exist. Technical measures can help: privacy-preserving computation techniques that enable AI analysis without exposing individual identities, data minimisation practices that collect only information necessary for specified purposes, and architectural choices that prevent the aggregation of data across systems into comprehensive citizen profiles. Governance measures are equally important: clear legal frameworks defining what surveillance is permitted and for what purposes, independent oversight with genuine authority and access, and meaningful transparency that enables public understanding and debate about the systems operating in their cities. World Economic Forum smart city principles provide one framework for responsible development, though implementation varies dramatically across jurisdictions. The MENA governments most committed to building trustworthy smart cities will be those that demonstrate governance sophistication alongside technological ambition—showing that surveillance infrastructure can be constrained by rule of law rather than merely deployed for state convenience.

Citizen Services and the Promise of Digital Government

AI-enabled citizen services represent a more unambiguously positive domain for government AI, offering efficiency improvements and accessibility enhancements that benefit both government operations and the citizens they serve. The UAE’s government services have achieved among the highest digital adoption rates globally, with platforms like UAEPASS providing unified digital identity that enables citizens and residents to access services across federal and local authorities. AI capabilities enhance these platforms through chatbots that handle routine enquiries, document processing systems that automate verification tasks, and recommendation engines that guide users toward services relevant to their circumstances. The UAE government’s happiness initiatives explicitly incorporate AI analysis of citizen feedback and service metrics, attempting to optimise government operations around citizen satisfaction outcomes rather than bureaucratic process measures.

Saudi Arabia’s digital government transformation has accelerated dramatically under Vision 2030, with the Tawakkalna app becoming ubiquitous during the pandemic and demonstrating the state’s capacity for rapid digital service deployment. The National Data Management Office has established the data infrastructure that enables AI-driven services, while initiatives like the Electronic Government Transactions Programme (Yesser) provide frameworks for digital service delivery across ministries. Citizen portals aggregate services that previously required separate interactions with multiple agencies, using AI to personalise interfaces and guide users through complex processes. The Kingdom has explicitly targeted reduction in government service delivery times as a key performance indicator, with AI automation central to achieving these targets. These improvements offer genuine value to citizens: reduced waiting times, elimination of unnecessary visits to government offices, faster processing of applications and requests, and more consistent application of rules and procedures.

The limits of AI in citizen services become apparent when examining complex, high-stakes decisions that require human judgement and accountability. While AI can appropriately handle routine transactions—processing a license renewal, scheduling an appointment, providing information about eligibility requirements—decisions affecting fundamental rights and life circumstances require human decision-makers who can be held accountable for their choices. OECD analysis of AI in government emphasises the importance of human oversight for consequential decisions, noting that automation of such decisions risks removing the accountability that legitimises state power. Immigration decisions, benefit determinations, law enforcement actions, and similar high-stakes governmental functions may appropriately use AI to support human decision-makers with information and analysis, but substituting algorithmic judgement for human accountability fundamentally changes the relationship between citizen and state. MENA governments developing AI capabilities for citizen services must distinguish between appropriate automation of routine processes and inappropriate removal of human judgement from decisions that affect fundamental rights.

Governance Challenges for Government AI

The governance of government AI presents distinctive challenges that public sector leaders across MENA must address. Unlike private sector AI, where market competition and regulatory oversight provide external checks on deployment, government AI often operates without equivalent accountability mechanisms. Governments both deploy AI and regulate its use, creating potential conflicts of interest when state interests in capability development conflict with citizen interests in protection from algorithmic harm. Responsible AI frameworks developed for private sector deployment require adaptation for government contexts, where the power differential between deploying organisation and affected individuals is even more pronounced than in commercial relationships. The citizens affected by government AI typically cannot choose alternative providers; they must interact with government services regardless of whether they trust the AI systems those services employ.

Procurement and acquisition of government AI systems raises particular concerns about transparency and accountability. When governments purchase AI capabilities from private vendors, the proprietary nature of those systems may prevent meaningful public understanding of how decisions affecting citizens are made. The “black box” problem—the difficulty of understanding how complex AI systems reach particular conclusions—becomes more acute when combined with commercial confidentiality claims that shield algorithmic details from public scrutiny. Some jurisdictions have begun requiring algorithmic transparency for government AI procurement, mandating that vendors provide sufficient documentation and explanation to enable meaningful oversight. Amsterdam’s algorithm register provides one model, publicly documenting AI systems used in city government and their purposes. MENA governments committed to accountable AI deployment should consider similar transparency requirements, recognising that public trust in government AI depends on public understanding of how these systems operate.

Building government capacity for AI governance requires investment in skills that public sectors have historically undervalued. Technical expertise to evaluate AI systems, understand their limitations, and oversee their operation remains scarce in government workforces worldwide. OECD guidance on digital government leadership emphasises the importance of developing internal government capacity rather than relying entirely on vendor expertise that may not align with public interest. MENA governments have recognised this need, with various initiatives to build AI skills within the public sector, though progress varies across agencies and jurisdictions. The governments that achieve the most from AI investment will be those that develop sophisticated internal capacity to specify requirements, evaluate proposals, oversee implementations, and hold systems accountable for outcomes—treating AI as a capability requiring ongoing governance rather than a procurement to be completed and forgotten.