From gratification to algorithmic resistance: The mediating role of information privacy concern in generative AI use

Abstract

Generative AI functions as an information service in which prompting can simultaneously seek information and disclose contextual data. Drawing primarily on privacy calculus, with uses-and-gratifications and algorithmic-resistance research specifying benefits and behavioural pathways, this study examines how information-seeking and social presence gratifications relate to adapted resistance-related practices through information privacy concern. We analysed survey data from 719 generative AI users in China using partial least squares structural equation modelling. Both gratifications were positively associated with privacy concern, which was positively associated with self-reported restriction (AAR) and signal-modifying/inference-disrupting (OAR) practices. Information privacy concern statistically accounted for the indirect association between information-seeking gratification and AAR and partially accounted for the other indirect associations. The findings position prompting as a dual information transaction while showing that OAR should not be treated as demonstrated privacy-protective resistance, because prompt variation may also reflect exploratory information seeking or strategic prompt engineering.