NDSS 2025 – Black-Box Membership Inference Attacks Against Fine-Tuned Diffusion Models

NDSS 2025 Session 12C features a paper by Yan Pang and Tianhao Wang introducing a scores-based, black-box membership inference attack framework against fine-tuned diffusion image generators, showing strong performance (AUC up to 0.95) across four scenarios and three attack types; this post syndicates the conference content and does not detail a specific security incident.