Work

AI-generated image-based sexual abuse: Perpetration and consumption across three regions

Year: 2026

Type: article

Abstract: The rapid pace of advancements in AI, paired with widespread availability and decreasing technical barriers, has resulted in significant concern about the generation of synthetic sexualized imagery (e.g., sexual ``deepfakes''). We surveyed a representative sample of 7,231 respondents in Australia, the United Kingdom, and the United States to map the prevalence of perpetrating and consuming AI-generated image-based sexual abuse (AI-IBSA). In those three regions, the overall population level rate of creating, sharing, and/or threatening to share images is 3.2%. Men, those under 35, BIPOC respondents, and respondents with a disability are significantly more likely than their counterparts to report these behaviors. 18% of all respondents report deliberately viewing sexual deepfakes, most commonly due to curiosity according to both men and women respondents. These findings suggest that, in addition to working to prevent the creation of nonconsensual AI-generated sexual images, sociotechnical interventions are needed to address the seeming normalization of consuming these images. Potential legal, technical, and educational interventions are discussed.

Source: Computers in Human Behavior

DOI: 10.1016/j.chb.2026.108935

Authors: Rebecca Umbach, Nicola Henry, Renee Shelby, Gemma Stevens, Kwynn Gonzalez-Pons

Institutions: Google (United States), RMIT University, YouTube (United States)

Language: English

Topic: Misinformation and Its Impacts

Subfield: Sociology and Political Science

Field: Social Sciences

Domain: Social Sciences

SDGs: Gender equality, Good health and well-being, Peace, justice, and strong institutions

Open Access status: hybrid

Estimated APC paid: $3,780