一作ACL 2026Poster· Natural Language Processing
DMN : A Compositional Framework for Jailbreaking Multimodal LLM s with Multi-Image Inputs
Wenzhuo Xu · Zhipeng Wei · Zonghao Ying · Deyue Zhang · Dongdong Yang · Xiangzheng Zhang · Quanchen Zou
摘要
Multimodal Large Language Models (MLLMs) are vulnerable to jailbreak attacks, which can elicit harmful responses from MLLMs. Many MLLMs support multi-image inputs, inadvertently introducing new vulnerabilities due to less efforts on multi-image safety alignment. Previous MLLM jailbreak methods on...