Generative AI and privacy: A protection motivation perspective : How do perceived data privacy threats and coping appraisals influence protection motivation in the context of Generative AI use?

Abstract

As the adoption of Generative AI (Generative Artificial Intelligence) accelerates across various domains, concerns about how these systems handle personal data are becoming increasingly relevant. This study investigates how individuals perceive data privacy threats associated with Generative AI and how these perceptions influence their motivation to adopt privacy-protective behaviors. Drawing on Protection Motivation Theory (PMT), six independent variables were examined: perceived threat severity, perceived threat vulnerability, maladaptive rewards, response efficacy, self-efficacy, and response costs. Additionally, the study explored whether self-rated awareness of AI-related data practices moderates the relationship between these predictors and protection motivation. Data were collected through a qualitative survey and analyzed using multiple linear regression. The results indicate that perceived threat, vulnerability, and response efficacy are positively associated with protection motivation, whereas response costs have a significant negative impact on protection motivation. However, perceived threat severity, maladaptive rewards, and self-efficacy were not significant predictors of the outcome. Moreover, self-rated awareness did not moderate any of the tested relationships, suggesting that general familiarity with AI data practices may not significantly alter behavioral intentions in this context. These findings offer both theoretical and practical contributions. They extend PMT to the domain of Generative AI privacy and highlight which psychological factors most strongly influence protective motivation. The study suggests that effective privacy communication strategies should focus on emphasizing vulnerability and the low effort required to act, rather than on abstract awareness-raising or general threat severity. Implications for developers, policymakers, and future research are discussed.