Aligning Urban AI and Global AI Governance: Insights from a Paris AI Action Summit Side Event

On February 11, 2025, The Governance Lab (The GovLab) and Urban AI co-hosted an official side event of the Paris AI Action Summit, titled "Aligning Urban AI and Global AI Governance." Held in collaboration with Mouvement des Entreprises de France (MEDEF), Open Data France, DemocracyNext, and UN-Habitat, the event brought together policymakers, researchers, and city representatives to discuss how urban AI initiatives can align with broader governance frameworks to ensure responsible and inclusive AI deployment. 

The rich and diverse lineup of speakers included Stefaan Verhulst, Co-Founder of The GovLab (NYC) and The DataTank (Brussels); Beatriz Botero, Assistant Professor of Law at Sciences Po; Hubert Beroche, Founder of Urban AI; Abdinassir Sagar, Programme Manager Officer at UN-Habitat; Constance Nebbula, Deputy Mayor of Angers Métropole; Mathijs Dewaele, Smart City Coordinator for the City of Oostende; Carl Morch, Co-Director of FARI – AI for the Common Good Institute at ULB Bruxelles; Claudia Chwalisz, Founder and CEO of DemocracyNext; Renee Sieber, Principal Investigator and Associate Professor at McGill University’s Department of Geography; Marie Bernard, Digital and Innovation Project Manager at Nantes Métropoles; Marlene Damerau, Head of the Smart City Division at gkd-el; and Gérard Wolf, Sustainable City “Federator” at the Ministry of Europe and Foreign Affairs, President of the Sustainable City Task Force at MEDEF International, Pierre Riou representing the Communauté d’Agglomération Paris-Saclay.

The Overlooked Role of Cities in AI Governance

Gérard Wolf, Sustainable City “Federator” at the Ministry of Europe and Foreign Affairs and President of the Sustainable City Task Force at MEDEF International, opened the event by stating that urban AI is essential for cities for three key reasons. First, urban migration is accelerating, with Africa's urban population expected to double by 2050, leading to critical challenges in managing infrastructure, resources, and informal settlements. Second, environmental issues demand sustainable AI solutions, as cities face increasing pressures on water, energy, and waste management. In this context, Wolf pointed to the example of Solideo, which constructed the Olympic Village in Paris with 47% lower carbon emissions than previous benchmarks. Lastly, he warned against overly centralized AI control, stating, "We do not want a Big Brother!"

Following Wolf, Dr. Stefaan Verhulst, Co-Founder of The GovLab (NYC) and The DataTank (Brussels), delved into the role of local governments in AI governance in his opening speech.  He pointed to both the advantages and challenges of this decentralized approach, as outlined in his presentation. AI localism, he argued, is a powerful but undervalued force in shaping responsible and inclusive AI governance. He also cited algorithmic transparency policies from cities like Amsterdam and Barcelona, which pioneered and served as models for broader-scale initiatives.

Cities at the forefront of AI Governance

Following the opening speeches, the first panel explored how cities are already leading AI governance further, often setting precedents for national and global policies. The discussants agreed that cities are not merely implementers of AI policies; they are active policymakers, shaping how AI is designed, deployed, and regulated within their jurisdictions. 

Mathijs Dewaele, Smart City Coordinator for the City of Oostende and Abdinassir Sagar, Programme Manager Officer at UN-Habitat, highlighted the capacity and cost challenges cities face in AI governance. Yet, they noted that local policies can shape how AI is designed, applied, and integrated into daily life through procurement strategies, regulatory innovations, and local decision-making. They also evoked Amsterdam’s algorithmic registries that promoted transparency as a good example. Moreover, the discussion underscored the bargaining power of cities as hosts of the essential infrastructure powering AI—like electricity grids and universities that cultivate talent— as well as active users of AI in public service.

Constance Nebbula, Deputy Mayor of Angers Métropole, touched upon the importance of governing data as a public resource, citing how mobility data from ride-sharing companies can be leveraged for urban planning. She also emphasized the potential of data altruism for public benefit. 

Lastly, Marlene Damerau, Head of the Smart City Division at gkd-el, introduced Germany’s AI Initiative for Cities, which rapidly gathered 130 AI project ideas, fostering collaboration between researchers and city officials to develop sustainable, city-driven AI solutions.

Experimenting with Participatory AI Governance

Collaboration has emerged as another key topic during the panel. Alliances such as Open Data France and the German Urban-KI initiative are transforming cities into vibrant hubs of exchange and collective learning. These networks enable urban centers to share best practices, coordinate policy efforts, and drive innovation that is both sustainable and ethically grounded. As cities become laboratories for experimentation—testing new models of public participation, transparency, and data stewardship—they are paving the way for a more responsive and inclusive approach to AI governance.

The discussants highlighted that as AI becomes increasingly integrated into public services, many cities are experimenting with participatory processes to involve residents in decision-making. They noted that AI governance should reflect the concerns and priorities of the communities affected, rather than being dictated solely by governments and technology companies. Examples included structured participatory models—such as public deliberations, workshops, and digital engagement tools—that allow communities to shape how AI is used in public services and urban planning, along with ethical frameworks for AI procurement to ensure local values are upheld. However, challenges remain, as most citizen engagement efforts are limited in scale and require stronger institutional support and accountability frameworks. The discussion underscored the need to scale up these participatory models and integrate them into broader governance structures.

European Regulations and Cities 

The second panel, featuring a fire-side chat between Beatriz Botero, Assistant Professor of Law at Sciences Po, and Carl Mörch, Co-Director of FARI – AI for the Common Good Institute (Université Libre de Bruxelles), examined the intersection of urban AI initiatives and European regulatory frameworks. 

Botero noted that under the EU AI Act, cities will be classified as "deployers" of AI systems, meaning they must conduct fundamental rights impact assessments. She argued that cities will set the global standard for AI compliance and serve as test cases for ethical AI implementation.

Mörch emphasized that scaling AI governance beyond experiments remains a major challenge. Reinforcing Mörch’s points, Marlene Damerau noted that AI governance is 20% technology and 80% organization. She shared an example from Germany, where AI was used to adjust traffic signals during flood risks, requiring extensive interdepartmental coordination. 

Lastly, Botero indicated that the cities hosting excellent universities have a special role in urban AI governance. As a response, Mörch’s brought up the gap between academic research and public benefit, using the example of Boston, where universities have budgets larger than some state governments but lack widespread community engagement. Following this discussion, the speakers also delved into the Draghi Report, which highlights how the US outpaces Europe in commercializing innovation. The discussion underscored the need to strengthen the pipeline from AI research to real-world applications in European cities. Notably, an attendee commented on the importance of retaining AI talent in European cities while embracing diversity as a strength of cities differing from the “national technology” strategies. 

Bridging Local and Global AI Narratives

In the third part of the event, Hubert Beroche, Founder of Urban AI, made a presentation on the geopolitical role of the cities.  He argued that the US and China have weaponized AI narratives, with the US companies promoting "smart city" solutions and China advancing the "safe city" narrative through companies like Huawei. On the other hand, local governments could help communities create their own narratives. Discussions highlighted that local governments, by virtue of their proximity to citizens, are uniquely positioned to guide the discourse on AI governance. They can ensure that public voices remain central, effectively counterbalancing digitalization narratives often dominated by private corporations. For instance, many cities are pioneering transparency and accountability mechanisms—such as algorithmic registries and ethical guidelines for AI procurement—that may serve as prototypes for global AI governance frameworks.  In addition, cities hosting major universities or large populations exercise significant geopolitical influence. Nonetheless, without more robust institutional channels for local governments to engage with national and global policymakers, the innovative practices initiated by the local governments struggle with financing and scaling up.

Followingly, the discussants elaborated the ways to include citizens in decision making processes. Renee Sieber, Principal Investigator and Associate Professor at McGill University’s Department of Geography, noted that public participation in AI governance is often limited to user satisfaction surveys, rather than true deliberation.

Marie Bernard, Digital and Innovation Project Manager at Nantes Métropoles, shared how Nantes has pioneered participatory AI governance through citizen workshops, AI coffee sessions, and dystopian scenario exercises. The city systematically analyzes public feedback to identify key concerns about AI, like agency, acceptability, accessibility, sustainability, and resilience. It shows how public perception, within a year, evolved from random, “panic” speeches to meaningful narratives.

Claudia Chwalisz, Founder and CEO of DemocracyNext, called for new mechanisms of AI governance, advocating for citizens’ assemblies and deliberative democracy models. She introduced the digital sortition system, explaining that it works in two stages: first, a random selection of citizens is invited to participate, and then a second lottery is conducted among those who accept the invitation. This process is followed by a rotation mechanism to ensure fresh perspectives and broad participation over time. However, she expressed skepticism about global AI assemblies, questioning their lack of enforcement mechanisms. She emphasized the need for local deliberation on AI’s role in essential services like healthcare.

Final Reflections

Before wrapping up the event, speakers and participants discussed in a free format and shared their experiences. Pierre Riou from Communauté d’Agglomération Paris-Saclay presented an urban AI project integrating startups to develop sustainable digital twins for urban planning.

A recurring theme was the need to ensure human oversight in AI-driven decision-making, reinforcing that AI should augment, rather than replace, human judgment. This was particularly emphasized in the context of AI-generated insights and information extraction. Renee Sieber, warned against over-reliance on AI for summarizing public consultations, as algorithmic biases can distort public sentiment.

The discussants mentioned how different large language models (LLMs) produce varying outputs, which raises questions about consistency and reliability in information retrieval. One of the speakers added a critical perspective, emphasizing that AI-generated outputs should never be taken at face value—it is important for policymakers to still engage in the fundamental act of reading and interpreting results to ensure accuracy and relevance. 

The discussion also addressed a common misconception: while generative AI (such as ChatGPT) has gained significant public and media attention, most cities do not actually rely on it for policy-related tasks. Instead, they rely on more transparent and structured AI models, such as rule-based systems and machine learning models, which are often sufficient for most municipal needs and provide greater explainability and control over decision-making processes. 

A central theme of the discussion was the need for stronger collaboration between local governments, national regulators, and international AI governance bodies. Finally, the event underscored that strengthening urban governance frameworks, fostering participatory decision-making, and upholding ethical oversight are essential to ensuring that AI serves the public good, not only in global strategies but in the everyday realities of urban life.