Risk Governance in Resilient Cities and Artificial Intelligence: A New Political Stakeholder?

English (OpenAlex)

Cities face increasingly complex risks, from climate extremes to pandemics, while deploying artificial intelligence (AI) across urban operations. This article asks whether AI should be treated as a non-human stakeholder in urban risk governance. Integrating risk governance, anticipatory governance, and Actor–Network Theory, we conduct a comparative qualitative analysis of three city cases: Calgary (climate resilience and infrastructure planning), Copenhagen (public-building energy management), and Kansas City (AI-enhanced mobility). Across cases, AI systems shape problem framing, prioritize interventions, and execute microdecisions in real time, shifting from instruments to delegates and, at times, co-governors. We propose an AI-as-actor spectrum anchored to core risk-governance tasks and derive indicators of “effective agency” (e.g., automated problem framing, resource-allocation authority, auditability/contestability). Findings show that AI’s influence scales with how early models enter the governance cycle, how tightly outputs are coupled to execution, and how contestable algorithmic “passage points” are. Policy implications include public algorithm registers, ex ante impact assessments, human-in-the-loop thresholds with escalation rights, decision logging, and periodic audits to safeguard transparency, accountability, and equity. Reconceptualizing governance to include non-human actants does not confer personhood on AI; it clarifies its embedded role and the institutional tools needed to govern it.