When conversational AI sells: objectivity disruption, persuasion knowledge and social-validation cues in sponsored AI recommendations

Purpose Drawing on persuasion knowledge theory and machine heuristic theory, this research examines separately how sponsorship in conversational AI recommendations influences consumer responses through perceived manipulation and perceived AI objectivity and whether social validation cues mitigate these effects. Design/methodology/approach Across one pre-test and four experiments (total N = 732), this research tests sponsorship effects, two mediators in separate models, social validation as a boundary condition, and the full model in a scripted, single-turn chat-style interface. Findings Across static screenshots and a scripted, single-turn interface, sponsorship increased perceived manipulation and reduced perceived AI objectivity. In separate mediation models, each mediator carried significant indirect effects on trust, engagement intention, and purchase intention. Social validation improved product-directed evaluations but did not consistently alleviate sponsorship-related concerns about the AI recommender or its recommendation process. Originality/value This research offers a dual-route account of sponsorship backfire and an exploratory attribute-matching proposition such that product-directed social-validation cues may improve product evaluations without resolving agent-directed concerns.