Governing AI innovation in local government: how trust and risk perceptions shape public support

Abstract

This study examines how citizens' experiences with public sector AI shape understanding, trust, and risk perceptions, and how these factors influence public support for responsible AI in local government contexts. Survey data from 1,183 respondents across Australia, Hong Kong, and Saudi Arabia were analysed using structural equation modelling. The framework links experience, perceived understanding, trust, perceived risks and benefits, attitudes toward urban AI and surveillance AI, and the prioritisation of responsible AI principles. Experience with local government AI significantly enhances perceived understanding, which in turn strengthens trust. Trust increases perceived benefits and favourable attitudes toward AI-enabled urban services. However, perceived risks remain a central driver of responsible AI prioritisation. Positive attitudes toward routine urban AI applications are also associated with greater openness to more contested surveillance AI applications. Cross-context differences indicate that these pathways vary according to governance setting, sample composition, and exposure to AI-enabled services. The findings are constrained by cross-sectional, self-reported data and differences in sampling strategies across contexts. Future research should use longitudinal designs and extend the model across additional institutional settings. Responsible implementation requires trust-building, public engagement, and proactive risk governance. Local governments can adopt an experience-first approach by beginning with visible urban AI services that build familiarity and demonstrate public value, while extensions to surveillance or other sensitive applications require stronger safeguards, transparency, and public justification. The findings inform the development of transparent, accountable, and citizen-centred AI governance, supporting socially acceptable and equitable urban innovation. The study advances a comparative, local government-centred model of public sector AI innovation by integrating experiential, cognitive, and perceptual dimensions. It offers novel insights into public support mechanisms for responsible local government AI.