Differential Vulnerability to AI-Generated Images: Age-Related Disparities in Recognition Accuracy Between Older Adults and Children

While artificial intelligence (AI) rapidly integrates into everyday life, little research addresses whether vulnerable populations like older adults can reliably distinguish AI-generated images from real ones. Intuitively, older adults may be less reliable in doing so because visual working memory and fusiform face area networks decline with age. In this study, 93 participants (35 children, 28 adolescents, and 30 older adults) completed an image-classification task. A 3 (age group) × 2 (image type: AI vs. real) × 4 (category: Transport, Humans, Plants, Animals) factorial design was used to analyze accuracy rates. Results showed that older adults underperformed both younger groups when identifying AI-generated humans images and real animal images, yet outperformed children in identifying real human images. Other image categories showed no significant age-group differences. These findings reveal older adults' unique vulnerability to AI-generated images, which suggests that age-inclusive safeguards and targeted interventions may be necessary.