Exposure to AI-Generated Content, Body Image, and Self-Esteem Among Social Media Users: The Mediating Role of Body Image

Abstract

The rapid proliferation of AI-generated images, AI beauty filters, virtual influencers, and digitally synthesized appearance content on social media has raised concern about its psychological impact on users. This study examined the relationships among exposure to AI-generated content (AIGC), body image, and self-esteem, and tested whether body image mediated the AIGC–self-esteem relationship. A quantitative, cross-sectional, correlational design was employed with 300 social media users aged 18–30 (56.7% female), recruited via convenience sampling. AIGC exposure was measured with a self-developed scale, body image with the Body Image States Scale, and self-esteem with the Rosenberg Self-Esteem Scale. Data were analyzed using descriptive statistics, Cronbach's alpha, Pearson correlation, multiple regression, and mediation analysis. AIGC exposure was significantly negatively correlated with self-esteem (r = –.46, p < .01) and body image (r = –.52, p < .01), while body image was significantly positively correlated with self-esteem (r = .58, p < .01). Regression analyses showed AIGC exposure explained 21.2% of the variance in self-esteem and 27.0% of the variance in body image, while body image explained 33.6% of the variance in self-esteem. Mediation analysis confirmed that body image partially mediated the relationship between AIGC exposure and self-esteem (indirect effect B = –0.214, 95% CI [–0.289, –0.146]), with a significant direct effect remaining (B = –0.205, p < .001). Findings indicate that AI-generated appearance content undermines self-esteem both directly and indirectly through diminished body satisfaction, underscoring the need for digital literacy initiatives, AI-content transparency, and body image-focused psychological interventions.