The Impact of AI-Personalized Recommendations on Online Consumer Behaviour: Trust, Privacy Concerns, Intrusiveness, and Purchase Decisions

AI-personalized recommendations are increasingly used in online shopping, but their relationship with consumer trust, privacy concerns, perceived intrusiveness, and purchase behaviour remains complex. This study examines these relationships using a cross-sectional dataset of 503 eligible online shoppers. Construct scores were analysed using descriptive statistics, internal-consistency assessment, standardised ordinary least-squares regressions, and case-bootstrap estimates. Personalization was positively associated with trust (β = .619) and perceived intrusiveness (β = .484). In the purchase-related model, personalization (β = .285), trust (β = .332), and perceived intrusiveness (β = .197) showed positive conditional associations. The personalization-by-privacy-concern interaction was negative (β ≈ −.149), contrary to the initial expectation. Bootstrap estimates indicated positive indirect associations through trust and intrusiveness; however, these are not evidence of causal mediation. Interpretation is constrained by weak internal consistency for four constructs, item-content overlap, cross-sectional self-report data, and unavailable information about the original sampling frame and fieldwork context. The findings are therefore exploratory and should not be treated as validated latent-variable or causal effects. The study highlights the need for clearer construct measurement when examining consumer responses to AI-personalized recommendations.