Black-box Membership Inference Attacks against Fine-tuned Diffusion Models
Yan Pang
Network and Distributed System Security (NDSS) Symposium 2025 · Day 3 · Membership Inference
This talk, presented by Yan Pang at the NDSS Symposium, delves into the critical area of data privacy concerning the rapidly evolving landscape of generative AI, specifically **diffusion models**. The core subject is the development and application of **black-box membership inference attacks (MIAs)** against **fine-tuned conditional diffusion models**. As diffusion models like Stable Diffusion achieve unprecedented capabilities in generating photorealistic images and even serving as game engines, their reliance on massive datasets – often comprising billions of images – raises significant privacy and copyright concerns. The talk highlights that traditional MIAs, designed for simpler generative adversarial networks (GANs) or variational autoencoders (VAEs), are inadequate for the complex, multi-step generation process of diffusion models.
AI review
Legitimate academic security research with a clean contribution: applying black-box MIA to fine-tuned conditional diffusion models using the model's own objective function to guide feature selection. Solid NDSS-tier paper presentation, but the talk itself is methodical and dry — this is a conference paper read aloud, not a security research talk that will stick with you.