Privacy-by-Design Virtual Influencers: Algorithmic Disclosure, Ownership Labels, and Consent UX
(A2026-131595)

AUTHORS

Tilen Pigac, Hong Kong Metropolitan University

ABSTRACT

Virtual influencers (VIs) are increasingly used in digital marketing but raise ethical concerns around authenticity, accountability, and consumer privacy. Prior research shows that anthropomorphism and personalization can enhance engagement yet also heighten privacy concerns, often creating backfire effects. This study investigates whether layered privacy-by-design interventions—algorithmic disclosure, ownership label clarity, and consent user experience (UX)—can mitigate these risks and strengthen persuasion outcomes. Drawing on Privacy Calculus Theory, the Persuasion Knowledge Model, and the Computers as Social Actors framework, we conducted a 2 × 2 × 2 × 2 online experiment (N = 812) in an Instagram-style environment. Results showed that algorithmic disclosure significantly reduced privacy concerns, ownership labels increased persuasion knowledge, and granular consent UX further lowered concerns. These mechanisms exerted indirect effects on trust, brand attitude, and purchase intention. Effects were strongest for highly anthropomorphic VIs and in privacy-sensitive categories such as healthcare and finance. The findings extend privacy calculus by showing that consumer paradoxes are malleable, enrich persuasion knowledge theory by positioning disclosure as value-enhancing, and identify anthropomorphism as a boundary condition. For managers, the results provide actionable guidance on using layered transparency and consent design to deploy VIs ethically and effectively.