Every Picture Tells a Dangerous Story: Memory-Augmented Multi-Agent Jailbreak Attacks on VLMs

ArXi:2604.12616v1 Announce Type: new The rapid evolution of Vision-Language Models (VLMs) has catalyzed unprecedented capabilities in artificial intelligence; however, this continuous modal expansion has inadvertently exposed a vastly broadened and unconstrained adversarial attack surface. Current multimodal jailbreak strategies primarily focus on surface-level pixel perturbations and typographic attacks or harmful images; however, they fail to engage with the complex semantic structures intrinsic to visual data.