Exposure to media images depicting ‘ideal’ bodies is a long-standing precursor to negative body image, and with advancements in artificial intelligence (AI) guiding creative media processes, it is necessary to explore the ways in which AI learns and generates ideal bodies. This study is a content analysis of images generated using three common AI platforms (Dall-E, MidJourney, Stable Diffusion) whereby comparisons across athlete and non-athlete images were examined. 48 images (50% athlete, 50% female) were generated using consistent prompts across AI platforms, which were then systematically analyzed for body image features. Comparisons across athlete and non-athlete images for females and males were conducted. Deductive coding suggested that the majority of images depicted low or very low body fat (athletes: 100%, non-athletes: 95.6%, p = .43 for group difference) and high muscularity (athletes: 91.7%, non-athletes: 37.5%, p < .001 for differences between the groups). Images of males were significantly more muscular (p < 0.05) and wearing less revealing clothing than females (p < 0.05). Among athletes, female images were coded as more objectified than males (p < 0.05). The results suggest that current body ideals represent unrealistic standards, specifically in their promotion of low body fat and greater than average muscularity in both groups, despite obvious differences in activity levels and sport type. The results also suggest that female athletes are more objectified than their male counterparts. This research offers insight on the body ideals perpetuated in the media which emphasize unrealistic standards that may damage mental well-being.