Institutional trustworthiness on public attitudes toward facial recognition technology: Evidence from U.S. policing

Abstract

This study examines the relationship between institutional trustworthiness and public acceptability of Artificial Intelligence (AI)-based Facial Recognition Technology (FRT) on its uses in five U.S. policing scenarios - public protests, large events, public streets, and two artificial scenarios representing public partial and high FRT acceptability. Drawing on a 2021 U.S. nationally representative survey (n = 4679), logit models demonstrate that two institutional trustworthiness dimensions - integrity and ability - affect public FRT acceptability. The effect of integrity is consistent across the five scenarios, and the Average Marginal Effects (AMEs) indicate that the effect is largest when the public assesses FRT acceptability in public protests - that presents the greatest privacy concern to citizens. This study contributes to the emerging literature on AI technology adoption in public entities through the institutional trustworthiness lens. It expands the application of the institutional trustworthiness framework into the local context of U.S. police adopting FRT and highlights contextual implications for public organizations adopting intrusive technologies. This study challenges a pervasive assumption of trading privacy for security in adopting FRT-like intrusive AI technologies in public services and recommends the ethical use of AI-based FRT in its operation, implementation, and development for public institutions, policymakers, and the AI industry.