The Effects of Generative AI Advertising Disclosure on Brand Authenticity Perceptions Purchase Intention and Consumer Skepticism Among Generation Z Consumers

The increasing application of generative artificial intelligence (GenAI) in digital advertising has introduced substantial changes in how brands develop promotional content, communicate product information, and establish relationships with consumers. However, explicitly disclosing AI involvement in advertisement production may influence perceived brand authenticity, purchase intention, and consumer skepticism, particularly among Generation Z consumers who demonstrate considerable familiarity with digitally mediated brand communication. This study proposes an integrated computational and behavioural framework for investigating the direct, indirect, and conditional effects of generative AI advertising disclosure on consumer responses. A novel Disclosure–Authenticity–Skepticism Predictive Model (DASPM) is developed to quantify how disclosure characteristics, including disclosure visibility, specificity, timing, and perceived transparency, influence brand authenticity perceptions and purchase intention through consumer skepticism. The proposed framework integrates structural equation modelling (SEM), moderated mediation analysis, and machine learning techniques to examine both theoretically specified causal pathways and complex predictive relationships. A randomized experimental survey involving Generation Z consumers aged 18–29 years is proposed, using matched advertisements assigned to three conditions: undisclosed AI-generated advertising, explicitly disclosed AI-generated advertising, and human-created advertising. Brand authenticity, consumer skepticism, advertising credibility, AI familiarity, and purchase intention are measured using validated multi-item Likert scales. The DASPM framework incorporates latent-variable estimation, interaction effects, nonlinear feature transformations, and predictive modelling, with its performance evaluated against conventional multiple linear regression, partial least squares structural equation modelling (PLS-SEM), Random Forest, and Extreme Gradient Boosting (XGBoost). Comparative evaluation will employ coefficient of determination (R²), root mean squared error (RMSE), mean absolute error (MAE), and cross-validated predictive performance, alongside bootstrapped indirect-effect estimates and measurement validity assessments. The results section is designed to present comparative performance graphs, conditional indirect-effect plots, disclosure-treatment comparisons, and feature-importance analyses to determine whether transparent AI disclosure strengthens consumer trust or activates skepticism that reduces purchase intention. The study will also investigate whether perceived brand authenticity mediates the disclosure–purchase intention relationship and whether AI literacy and prior brand familiarity moderate these effects. The proposed framework is intended to advance understanding of generative AI advertising disclosure by combining experimental consumer research with interpretable predictive analytics, providing an evidence-based approach to evaluating disclosure strategies and their implications for brand communication among Generation Z consumers.