Visual Contextual Attack: Jailbreaking MLLMs with Image-Driven Context Injection.

Ziqi Miao, Yi Ding, Lijun Li, Jing Shao

Browse the full EMNLP paper archive.