Artificial intelligence driven approaches for responsible urban water management

Urban water systems are under growing stress from rising demand, aging infrastructure, and climate variability. Artificial Intelligence (AI) is increasingly promoted as a tool to make these systems more adaptive and efficient. This work focuses on three core applications in water distribution networks: pump scheduling, smart metering, and leak detection, highlighting their demonstrated potential to reduce energy use, cut non-revenue water, and improve demand forecasting. At the same time, the paper identifies critical operational barriers, including data scarcity, limited model generalizability, and the integration of AI with legacy systems, as well as ethical concerns around privacy, fairness, transparency, and automation bias. To bridge opportunities and risks, we propose a framework for responsible AI adoption that links technical applications to enabling conditions and governance needs. We argue for cautious optimism: when embedded within robust data infrastructures and coupled with human oversight, AI tools serve to make urban water management more resilient, trustworthy and efficient. We argue that while AI holds significant promise for transforming urban water management, their potential remains largely untapped in practice. Limitations include uneven data availability in water-scarce regions, lack of standardized protocols, and unsolved ethical concerns around data ownership and gaps around legislation. This article advocates for a more ethically grounded and systems-level integration of these technologies in future urban water strategies.