Deceptively similar: Understanding the role of ethnic affinity targeting in Gen-AI-created political advertising

Mengqi (Maggie) Liao and Joshua Cloudy, “Deceptively similar: Understanding the role of ethnic affinity targeting in Gen-AI-created political advertising.” Paper to be presented at the annual conference of the American Academy of Advertising, Austin, March 26-29, 2026. Abstract: Ethnic affinity targeting (EAT), tailoring advertising content to match an audience’s ethnicity, has long been shown to enhance persuasion, but its use in the era of generative AI (Gen-AI) raises new ethical and psychological questions. With Gen-AI making it easier to produce multiple ethnicity-matched versions of the same ad, it remains unclear whether traditional matching effects persist when the spokesperson is AI-generated, and whether lay users can accurately detect the AI-generated content. Guided by the personalized matching framework, we conducted a 2 (participant race: White vs. Black) × 2 (avatar race: White vs. Black) online experiment (N=201) to examine the impact of AI-generated avatars in political advertising. Results showed an ethnicity-matching effect among Black participants such that a Black AI spokesperson elicited higher perceived homophily, which in turn predicted stronger advocacy intentions. Notably, most participants failed to recognize that the spokesperson was AI-generated, suggesting low detection ability. These findings offer theoretical insights into EAT, and highlight critical practical implications for the ethical design and deployment of Gen-AI in political advertising.