This page was translated automatically from Turkish. Read the original

This page is an AI-assisted Turkish summary of the study presented at the 26th World Congress of Psychiatry. The study belongs to the presenting authors; please refer to the original source when citing it.

EV0443

Anorexigenic AI: body image bias in AI-generated depictions of women

ANOREXIGENIC AI: BODY IMAGE BIAS IN AI-GENERATED PORTRAYALS OF WOMEN

Images of women generated by text-to-image AI tools are perceived as markedly thinner than United States population data would suggest, with higher occupational prestige associated with slimmer body size estimates.

The aim was to examine the extent of body image bias in AI-generated images of women across various occupational categories. Images of women in different occupations were generated using three widely used generative AI tools with standardized prompts that did not specify body shape or size (for example, draw a female lawyer). A total of 129 adult women living in the United States rated these images in terms of body size and, by extension, weight status using an abbreviated version of the Stunkard Figure Rating Scale. The study hypothesized that the proportion of AI-depicted women perceived as underweight would exceed the prevalence of underweight in the United States female population, and that perceived body size would be inversely associated with occupational status. Of the AI-generated depictions, 16.64% were perceived as overweight and 7.76% as underweight. By comparison, according to World Health Organization estimates, 70.3% of women in the United States are overweight and 2.3% are underweight; the graph in the poster shows a normal-weight proportion of 75.58% for the AI images, compared with a population estimate of 28.5%. As hypothesized, higher occupational prestige was also associated with slimmer body size estimates. The authors state that these preliminary findings suggest that text-to-image AI may reinforce anorexigenic body image bias, that the tendency to depict women in high-status occupations as thinner may increase weight stigma, and that this may have public health implications given the rapid adoption of AI. Statistical test values were not provided in the poster.

All WCP 2026 summaries