Facial recognition technologies (FRT) have been widely studied across technical, ethical, and application-specific domains, yet their real-world adoption continues to raise persistent concerns about trust, governance, and responsible use. While prior systematic reviews have examined FRT from algorithmic, bias-related, or adoption-intention perspectives within specific use cases, they largely treat governance as contextual background and conceptualize trust and acceptance as static outcomes. This review addresses this gap by synthesizing the literature through a human–AI collaboration lens, explicitly integrating deployment environments, governance arrangements, and cognitive mechanisms such as trust calibration, vigilance, and automation bias. Based on a PRISMA-aligned systematic review of studies across public, commercial, and critical-infrastructure contexts, we propose a parsimonious integrative conceptual model that explains how regulatory design and institutional context shape not only acceptance but patterns of reliance, oversight, and epistemic risk during system use. By foregrounding cognition and governance as central mechanisms rather than peripheral factors, this review advances the extant literature beyond adoption-centered frameworks and offers a clearer theoretical foundation for understanding when and why FRT deployment leads to responsible human–AI collaboration or problematic overreliance. The findings provide actionable insights for researchers, policymakers, and practitioners seeking to design and regulate FRT systems in ways that support calibrated trust and mitigate unintended harms.
Bhatia et al. (Thu,) studied this question.