When Personalization Becomes Persuasion: Disclosure, Autonomy, and Conversion in Generative-AI Advertising

Generative AI enables advertisers to customize content on a per-person basis with minimal additional expense. The ability to make things seem more relevant can also diminish the persuasive message and reduce consumer choice. This study investigates the influence of personalization intensity and disclosure of AI generated content on the perceived relevance, manipulation, autonomy, trust, and conversion intention. A simulation-based 2 × 2 online experiment was designed using 1,248 synthetic respondent records, half assigned to disclosure-present and half to disclosure-absent settings, with half of the records in each of these disclosure conditions being assigned to low-personalization and the other half to high-personalization settings. Multi-item measures were analyzed by means of factorial analysis of variance, robust regression and 5,000-sample bootstrap mediation. High personalization also led to greater manipulation (M = 4.48) and less autonomy (M = 4.18) without disclosure, and greater relevance (M = 5.15) with disclosure (p < .001). Disclosure raised conversion under high personalization from M = 3.75 to M = 4.28, interaction F(1, 1244) = 17.80, p < .001, partial η² = .014. Conditional indirect effect pathway was mediated through autonomy, manipulation and trust was b = .48, 95% CI [.38, .58]. The results suggest that disclosure can be a governance tool supporting the conversion process if it can regain informed choice, rather than simply marking content. Personalization design can be made transparent to match performance with consumer agency.