Paper Published in Big Data & Society

Created on July 28, 2026

2026

Journal: Big Data & Society (SAGE), Vol. 13, Issue 3 Co-author: Eva Cetinić DOI: 10.1177/20539517261467361

Perpetuating Misogyny with Generative AI: How Model Personalization Normalizes Gendered Harm

Our paper analyzing CivitAI — the most active platform for sharing and developing open-source text-to-image models — is now published in Big Data & Society, after previously circulating as a preprint on arXiv.

Drawing on a dataset of more than 40 million user-generated images and over 230,000 models, we find a disproportionate rise in not-safe-for-work content and a significant number of models intended to mimic real individuals without consent. We contextualize these findings through feminist and constructivist perspectives on technology and propose interventions to mitigate downstream harm.

See the publications page for the full entry and abstract.