The rhythm of city life runs on thousands of quiet decisions that support citizens: how patterns in traffic lights limit traffic jams, how energy grids respond to rising demand to balance power needs, and where safety data dictates where police patrol.
Across the world, cities are quietly delegating the details of daily life to new artificial intelligence (AI) systems that make data driven recommendations about how resources and priorities are distributed. What begins as a promise of efficiency can also quietly define the invisible architecture of who controls public life.
Smart cities, constrained consumers
More than half of smart city projects worldwide now rely on AI, yet most are delivered through closed, vendor-owned systems. When cities sign contracts for algorithmic services, they rarely retain the rights to the data, the models, or the learning that these systems produce, which in turn limits their ongoing ability to govern or adapt the systems they deploy. As a result, smart cities risk outsourcing critical AI decisions to private suppliers.
Treating public AI solely as a procurement question is a missed opportunity. By the time a city finishes bidding for an AI system, technology has often moved on, leaving taxpayers to fund yesterday’s code. Moreover, public procurement processes often prioritise vendor stability and low costs over iterative, innovative solutions. While this purchasing logic may work for parking meters, AI is not static infrastructure. It evolves through use, feedback and civic oversight.
Seeing AI as a product to purchase also risks entrenching inequality and widening the divide between cities. While wealthier cities can commission bespoke, sophisticated AI systems, smaller municipalities will depend on generic, cloud-based tools that limit autonomy and the chance for exponential growth. Most of all, seeing AI as something to buy rather than build reduces cities to consumers with limited agency, rather than architects of their own digital future.
Governance gaps are leaving locals behind
For cities to act with agency on the global stage and remain accountable to their residents, they need to weigh AI development alongside housing, water, and other local priorities rather than absorb the consequences of outside decisions. AI development affects city residents beyond just digitised government service delivery; cities have a stake across the entire AI lifecycle. For example, in Newton County, Georgia, USA residents have watched their water wells run dry as new data centers move in town.
“It feels like we’re fighting an unwinnable battle we never signed up for,” one resident said as her tap ran dry, costs rose, and the county water commission warned of shortages.
This is not an inevitable outcome of AI growth; it reflects a governance gap between global technology ambitions, federal policy and the local communities left to live with the results. City-led AI infrastructure offers one way to close that gap and align technological growth with civic priorities.
When cities treat AI as a civic practice, they begin to build systems that reflect their values and strengthen public trust rather than products to support public function. Municipalities are well-positioned to lead by example in agile, experimental and accountable governance, as they tend to be more trusted than national governments and provide visible, responsive public services.
Across global smart city initiatives, the most effective AI use cases reported appear in public safety, energy management, urban planning and decision-making processes. These aren’t technical domains, but instead are core civic systems of social life. Each with its own tradition of community organising, coalition building and citizen participation.
By grounding AI development in strong existing civic mechanisms of public feedback, transparent decision-making, and local collaboration, cities can build infrastructure that not only serves communities but is continuously shaped by them.
Share of active AI use by city domain
How cities can move from consumers to co-creators
Emerging regional models hint at the opportunities of a new generation of civic infrastructure: intelligence that is distributed and designed for shared benefit. In Japan, the AI Bridging Cloud Infrastructure (ABCI) provides shared compute resources for smaller institutions and municipalities to access and build their own AI solutions.
India’s Data Empowerment and Protection Architecture (DEPA) is the nation’s model of data for public benefit while preserving each individual’s privacy through a federated data infrastructure. Meanwhile, the Netherlands has embraced multi-city AI alliances, like the Dutch Metropolitan Innovations initiative to share not just the data tools, but the institutional learnings and best practices developed across Dutch cities.
Each of these efforts demonstrates a different infrastructural path toward inclusive growth, from democratising compute to unlocking data for public benefit and sharing tools and knowledge across cities.
These efforts hint at the opportunities of technical capacity aligned with civic purpose. But without the institutional and human capacity to match, these models cannot mature into lasting civic infrastructure.
In 2021, Vienna’s officials cited a lack of skills as their biggest challenge in implementing its data strategy. Municipal leaders need the internal ability to design systems, set public standards and sustain ongoing maintenance. That means investing in new forms of governance and expertise, such as appointing chief algorithm officers, supporting citizen data trusts, and maintaining algorithm registries akin to planning commissions or public utilities.
AI for the public good
The work ahead calls for civic imagination and for ambitious leaders who can link algorithms to accountability and build systems that reflect human values as much as computational precision. AI as an infrastructure can enable capacity, models and governance frameworks to circulate across networks of cities, and provide a new chance for interoperable, collective growth.
In the last century, cities built roads, grids and waterworks. In the next, they must build the infrastructure of civic intelligence – the public architecture that determines who benefits from technology and who gets left behind.
Learn more about the OECD’s work on AI and Smart Cities: The OECD Programme on Smart Cities and Inclusive Growth. Be part of the conversation: OECD Roundtable on Smart Cities and Inclusive Growth.
AnissaArakal is a technology researcher and governance analyst examining how emerging technologies can serve human values. She has coordinated international dialogues and authored research on AI, digital public infrastructure, and cybersecurity. Her earlier work advanced human rights and privacy frameworks for the design and governance of facial recognition technologies. She holds a B.A. in Politics, Technology, and Society from Barnard College, Columbia University.
Vilas Dharis a globally recognised authority on artificial intelligence and society, serving as President of thePatrick J. McGovern Foundation, a $1.5 billion philanthropy advancing AI for public purpose. He has served on theUnited Nations Secretary-General’s High-Level Advisory Bodyon Artificial Intelligence, is the U.S. Government’s nominated expert to theGlobal Partnership on AI, and advisesStanford’s Institute for Human-CenteredAI, OECD.AI,andMIT Solve. Named aWorld Economic Forum Young Global Leaderin 2022, he is also a public intellectual whose writing and teaching have reached audiences worldwide, including 500,000 learners of hisLinkedIn Learning course.