Saved in:
Bibliographic Details
Main Authors: Yu, Mingyu, Liu, Lana, Zhao, Zhehao, Wang, Wei, Qin, Sujuan
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2601.15698
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917217181892608
author Yu, Mingyu
Liu, Lana
Zhao, Zhehao
Wang, Wei
Qin, Sujuan
author_facet Yu, Mingyu
Liu, Lana
Zhao, Zhehao
Wang, Wei
Qin, Sujuan
contents The rapid advancement of Multimodal Large Language Models (MLLMs) has introduced complex security challenges, particularly at the intersection of textual and visual safety. While existing schemes have explored the security vulnerabilities of MLLMs, the investigation into their visual safety boundaries remains insufficient. In this paper, we propose Beyond Visual Safety (BVS), a novel image-text pair jailbreaking framework specifically designed to probe the visual safety boundaries of MLLMs. BVS employs a "reconstruction-then-generation" strategy, leveraging neutralized visual splicing and inductive recomposition to decouple malicious intent from raw inputs, thereby leading MLLMs to be induced into generating harmful images. Experimental results demonstrate that BVS achieves a remarkable jailbreak success rate of 98.21\% against GPT-5 (12 January 2026 release). Our findings expose critical vulnerabilities in the visual safety alignment of current MLLMs.
format Preprint
id arxiv_https___arxiv_org_abs_2601_15698
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Beyond Visual Safety: Jailbreaking Multimodal Large Language Models for Harmful Image Generation via Semantic-Agnostic Inputs
Yu, Mingyu
Liu, Lana
Zhao, Zhehao
Wang, Wei
Qin, Sujuan
Computer Vision and Pattern Recognition
Artificial Intelligence
The rapid advancement of Multimodal Large Language Models (MLLMs) has introduced complex security challenges, particularly at the intersection of textual and visual safety. While existing schemes have explored the security vulnerabilities of MLLMs, the investigation into their visual safety boundaries remains insufficient. In this paper, we propose Beyond Visual Safety (BVS), a novel image-text pair jailbreaking framework specifically designed to probe the visual safety boundaries of MLLMs. BVS employs a "reconstruction-then-generation" strategy, leveraging neutralized visual splicing and inductive recomposition to decouple malicious intent from raw inputs, thereby leading MLLMs to be induced into generating harmful images. Experimental results demonstrate that BVS achieves a remarkable jailbreak success rate of 98.21\% against GPT-5 (12 January 2026 release). Our findings expose critical vulnerabilities in the visual safety alignment of current MLLMs.
title Beyond Visual Safety: Jailbreaking Multimodal Large Language Models for Harmful Image Generation via Semantic-Agnostic Inputs
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2601.15698