When Nobody Knows What Is Real: Associations Between AI-Generated Manipulation, Fear of AI Fraud, Privacy Protection Behavior, and Trust in AI Governance

Abstract

The rapid proliferation of generative artificial intelligence (AI) has intensified concerns regarding misinformation, identity fraud, privacy, and public trust in AI governance. This study examines the associations among perceived exposure to AI-generated manipulation, fear of AI fraud, privacy protection behavior, and trust in AI governance. Informed by Protection Motivation Theory, a theoretically specified sequential model of associations among Exposure to AI Manipulation, Fear of AI Fraud, Privacy Protection Behavior, and Trust in AI Governance was evaluated. Data were collected from 2683 respondents across ten countries using Prolific. The measurement model was assessed through exploratory and confirmatory factor analyses, while the proposed associations were examined using covariance-based structural equation modeling and bootstrap analysis of specific indirect associations. Perceived exposure to AI manipulation was positively associated with fear of AI fraud, which was positively associated with privacy protection behavior. Privacy protection behavior was positively associated with trust in AI governance. However, exposure was not directly associated with privacy protection behavior or trust, while fear was not directly associated with trust. Significant specific indirect and sequential indirect associations were identified within the specified model and should be interpreted as model-dependent statistical associations. Interpreted through the lens of Protection Motivation Theory, the findings identify a theoretically consistent pattern of associations among perceived technological exposure, AI-related fear, privacy-protective behavior, and institutional trust. Because the data are cross-sectional, these associations do not establish temporal or causal ordering among the constructs.