UN Warns Cities: Keep AI Under Human Control

The United Nations says city authorities must be ready to reject or discontinue artificial-intelligence systems that threaten the public interest, undermine human rights or introduce unacceptable risks. The guidance targets decisions made through public services and administrative systems—decisions that affect residents, while responsibility can disappear behind an automated process. The UN’s answer: algorithms must not become the final authority.

Who Controls the Decision

The guidance was developed by the International Telecommunication Union (ITU) and the UN Human Settlements Programme (UN-Habitat). Its four principles tell city governments to maintain meaningful human control over decisions, assess potential harm, prioritise actual urban challenges over technology adoption, and stop an AI application when risks outweigh benefits or adequate safeguards can’t be assured. The stop button is part of the policy, not an admission that every new technology deserves a place in public life.

The United for Smart Sustainable Cities (U4SSC) Toolbox for People-Centred Smart Cities in the AI Era covers municipal systems used to manage public services, respond to residents’ complaints, identify infrastructure problems and support administrative decisions. Those are functions through which city authorities shape people’s access to services and the handling of their concerns. The UN warns that poorly governed systems could introduce bias, compromise privacy, undermine human rights and create accountability gaps when automated decisions affect residents.

City authorities are advised to define the problem they intend to solve, consider who could benefit or suffer, assess whether deployment is justified, and strengthen safeguards for applications with higher consequences. They’re also told to establish monitoring and accountability, then decide when systems should be modified or discontinued. In other words, buying or deploying a system doesn’t settle the question of whether it should govern people’s lives.

The People Missing From the Data

Anacláudia Rossbach, UN-Habitat executive director, said AI and digital innovation must respond to community needs, particularly those of residents in informal settlements and underserved areas. She identified gaps in data, human rights protections and meaningful participation by residents and civil society as important considerations in decisions about urban technology. The guidance recognizes participation as a concern; the article describes no resident-led decision-making process or community response to the toolbox.

The stakes are especially pronounced in developing economies, where governments are looking to digital platforms, data infrastructure and AI to address urban challenges while access to reliable data, infrastructure and digital skills remains uneven. The source warns that communities poorly represented in the data used to train or operate automated systems could face existing inequalities reproduced in public service delivery. Those with less representation in the data are also among those whose lives these systems may influence.

The toolbox draws on smart-city experiences in Helsinki, Finland, and eThekwini, South Africa. Doreen Bogdan-Martin, ITU secretary-general, said the resource would help city officials translate international standards and responsible AI principles into practical guidance for improving public services and building public trust. That is the institutional promise. The guidance also calls for clear accountability structures before AI enters public infrastructure and administrative processes.

The Apparatus Gets a New Toolkit

The launch coincided with the 10th anniversary of United for Smart Sustainable Cities, a global initiative coordinated by ITU, UN-Habitat and the UN Economic Commission for Europe, with support from 19 other UN agencies and programmes. Established in 2016, U4SSC provides tools, methodologies and performance indicators to help cities accelerate sustainable digital transformation and advance the United Nations Sustainable Development Goals.

Alongside the AI toolbox, the initiative released publications on urban AI governance, measuring people-centred digital transformation, security for digital public infrastructure, city data platforms, anticipatory governance and data-enabled destination management. The publications are intended to give city and national authorities a common framework for using AI and data to build more inclusive, resilient and responsive urban environments. The guidance sets limits and demands oversight, but the authority to deploy, monitor or discontinue these systems remains with city institutions.