The AI Disclosure Paradox in Marketing: A Persuasion Knowledge Authenticity Framework
Main Article Content
Abstract
Regulators and industry codes increasingly require brands to disclose when consumer-facing content or interactions involve artificial intelligence, yet a growing body of marketing evidence shows that disclosure itself carries a behavioral cost: it can activate consumer skepticism, reduce perceived authenticity, and depress trust and purchase intention. This conceptual paper integrates the persuasion knowledge model (Friestad & Wright, 1994), theories of consumption authenticity (Beverland & Farrelly, 2010), algorithm aversion research (Dietvorst, Simmons, & Massey, 2015), and recent empirical evidence on AI chatbot and AI-generated advertising disclosure (Luo, Tong, Fang, & Qu, 2019; Longoni & Cian, 2022; Kučinskas & Survilaitė,) to develop a Persuasion Knowledge–Authenticity Framework explaining when AI disclosure helps versus harms marketing outcomes. The framework specifies five interacting constructs — disclosure salience, task-type congruence, perceived authenticity threat, persuasion knowledge activation, and track-record sensitivity — and advances six propositions describing how these constructs jointly determine the size and direction of disclosure effects on consumer trust and purchase intention. The paper contributes to marketing theory by reframing AI transparency as a contingent design problem rather than a uniformly positive or negative practice, and it offers marketers a diagnostic vocabulary for calibrating disclosure strategy across utilitarian and hedonic contexts, brand familiarity levels, and consumer AI-literacy segments. To illustrate the framework's plausibility ahead of purpose-built testing, the paper reanalyzes published results from a nationally representative U.S. survey (Pew Research Center's American Trends Panel, June, N = 5,023) covering seven AI-disclosure vignettes, finding a pattern of disclosure penalties broadly consistent with the framework's emphasis on persuasion-relevant and consequential contexts. Boundary conditions, managerial implications, and a future research agenda are discussed.