Efficient Generation of Targeted and Transferable Adversarial Examples for Vision-Language Models Via Diffusion Models https://
arxiv.org/abs/2404.10335

→ View original post on X — @jiqizhixin