Dark Side of Artificial Intelligence: A Meta‐Analytic Review

ABSTRACT The rapid integration of artificial intelligence (AI) into consumer interactions has sparked concerns about its darker implications. When AI‐enabled service encounters are perceived as error‐prone, socially disruptive, emotionally misattuned, privacy‐invasive, or ethically questionable, consumers may develop distrust and resistance towards adoption. Despite growing scholarly attention, research on AI‐related risks and consumer resistance remains fragmented and empirically inconsistent, limiting clarity on which risk dimensions most consistently generate distrust, how strongly distrust translates into resistance, and why effect sizes vary across studies and contexts. To address this gap, we conducted a meta‐analysis grounded in innovation resistance theory and the theory of distrust by synthesizing findings from 30 empirical studies encompassing 302 independent effect sizes and 109,758 participants. We estimated the pooled effects of performance, social, emotional, privacy and security, and ethical risks on distrust and tested the distrust‐to‐resistance pathway using meta‐analytic structural equation modeling. The findings reveal privacy and security risk and ethical risk emerge as the strongest predictors of distrust, which in turn fuels resistance behavior. Further, moderator analyses explain prior inconsistencies by showing that these relationships systematically vary across gender composition, economic development, cultural orientations, and digital readiness. Specifically, social and emotional risks are more influential in developing economies and collectivist cultures, whereas ethical concerns are more salient in digitally mature regions. This study contributes to the ongoing discourse on AI ethics, trust, and adoption by consolidating fragmented evidence into a coherent framework, clarifying why and when consumers resist AI in services, and offering actionable guidance for AI developers, policymakers, and marketers seeking to advance trust‐centered adoption strategies and context‐sensitive AI design.