How does risk information exposure affect AI-driven technology users’ privacy protection? Combining social amplification of risk framework and technology threat avoidance theory

The pervasive proliferation and integration of AI-driven technologies across myriad domains shifts individual lifestyles and organizational productivity, while simultaneously catalyzing the privacy risk within ICTs. Although the narratives surrounding the privacy risk associated with AI-driven technologies have begun to emerge in public discourse, the consequences of exposure to such information remain unexplored. Combining the social amplification of risk framework (SARF) and technology threat avoidance theory (TTAT), the current study constitutes an exploratory empirical framework focusing on the association between risk information exposure and users' privacy protection behavior. Empirical data were collected from 787 participants who had adopted AI-driven technologies via the online survey platforms Credamo and Sojump. Data was examined by the partial least squares structural equation modeling (PLS-SEM) approach. The outcomes confirmed that risk information exposure was positively correlated with both perceived susceptibility and perceived severity, which were positively correlated with privacy protection motivation. Moreover, safeguard effectiveness and self-efficacy were positively associated with privacy protection motivation, while safeguard cost did not demonstrate a significant association. Privacy protection motivation exhibited a positive association with privacy protection behavior, and this relationship was positively moderated by information security culture. Consequently, this research advances the comprehension of risk communication related to AI-driven technologies and the dynamic mechanisms underlying users’ privacy protection behavior within the unique socio-cultural environment of China. It simultaneously offers pragmatic recommendations for stakeholders such as policymakers, AI-driven technology developers, and media practitioners to solve the privacy risk within the AI-mediated environment.