OmniSafeBench-MM: A Unified Benchmark and Toolbox for Multimodal Jailbreak Attack-Defense Evaluation

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Jia, Xiaojun, Liao, Jie, Guo, Qi, Ma, Teng, Qin, Simeng, Duan, Ranjie, Li, Tianlin, Huang, Yihao, Zeng, Zhitao, Wu, Dongxian, Li, Yiming, Ren, Wenqi, Cao, Xiaochun, Liu, Yang
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866917131401035776
author Jia, Xiaojun
Liao, Jie
Guo, Qi
Ma, Teng
Qin, Simeng
Duan, Ranjie
Li, Tianlin
Huang, Yihao
Zeng, Zhitao
Wu, Dongxian
Li, Yiming
Ren, Wenqi
Cao, Xiaochun
Liu, Yang
author_facet Jia, Xiaojun
Liao, Jie
Guo, Qi
Ma, Teng
Qin, Simeng
Duan, Ranjie
Li, Tianlin
Huang, Yihao
Zeng, Zhitao
Wu, Dongxian
Li, Yiming
Ren, Wenqi
Cao, Xiaochun
Liu, Yang
contents Recent advances in multi-modal large language models (MLLMs) have enabled unified perception-reasoning capabilities, yet these systems remain highly vulnerable to jailbreak attacks that bypass safety alignment and induce harmful behaviors. Existing benchmarks such as JailBreakV-28K, MM-SafetyBench, and HADES provide valuable insights into multi-modal vulnerabilities, but they typically focus on limited attack scenarios, lack standardized defense evaluation, and offer no unified, reproducible toolbox. To address these gaps, we introduce OmniSafeBench-MM, which is a comprehensive toolbox for multi-modal jailbreak attack-defense evaluation. OmniSafeBench-MM integrates 13 representative attack methods, 15 defense strategies, and a diverse dataset spanning 9 major risk domains and 50 fine-grained categories, structured across consultative, imperative, and declarative inquiry types to reflect realistic user intentions. Beyond data coverage, it establishes a three-dimensional evaluation protocol measuring (1) harmfulness, distinguished by a granular, multi-level scale ranging from low-impact individual harm to catastrophic societal threats, (2) intent alignment between responses and queries, and (3) response detail level, enabling nuanced safety-utility analysis. We conduct extensive experiments on 10 open-source and 8 closed-source MLLMs to reveal their vulnerability to multi-modal jailbreak. By unifying data, methodology, and evaluation into an open-source, reproducible platform, OmniSafeBench-MM provides a standardized foundation for future research. The code is released at https://github.com/jiaxiaojunQAQ/OmniSafeBench-MM.
format Preprint
id arxiv_https___arxiv_org_abs_2512_06589
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle OmniSafeBench-MM: A Unified Benchmark and Toolbox for Multimodal Jailbreak Attack-Defense Evaluation
Jia, Xiaojun
Liao, Jie
Guo, Qi
Ma, Teng
Qin, Simeng
Duan, Ranjie
Li, Tianlin
Huang, Yihao
Zeng, Zhitao
Wu, Dongxian
Li, Yiming
Ren, Wenqi
Cao, Xiaochun
Liu, Yang
Cryptography and Security
Computer Vision and Pattern Recognition
Recent advances in multi-modal large language models (MLLMs) have enabled unified perception-reasoning capabilities, yet these systems remain highly vulnerable to jailbreak attacks that bypass safety alignment and induce harmful behaviors. Existing benchmarks such as JailBreakV-28K, MM-SafetyBench, and HADES provide valuable insights into multi-modal vulnerabilities, but they typically focus on limited attack scenarios, lack standardized defense evaluation, and offer no unified, reproducible toolbox. To address these gaps, we introduce OmniSafeBench-MM, which is a comprehensive toolbox for multi-modal jailbreak attack-defense evaluation. OmniSafeBench-MM integrates 13 representative attack methods, 15 defense strategies, and a diverse dataset spanning 9 major risk domains and 50 fine-grained categories, structured across consultative, imperative, and declarative inquiry types to reflect realistic user intentions. Beyond data coverage, it establishes a three-dimensional evaluation protocol measuring (1) harmfulness, distinguished by a granular, multi-level scale ranging from low-impact individual harm to catastrophic societal threats, (2) intent alignment between responses and queries, and (3) response detail level, enabling nuanced safety-utility analysis. We conduct extensive experiments on 10 open-source and 8 closed-source MLLMs to reveal their vulnerability to multi-modal jailbreak. By unifying data, methodology, and evaluation into an open-source, reproducible platform, OmniSafeBench-MM provides a standardized foundation for future research. The code is released at https://github.com/jiaxiaojunQAQ/OmniSafeBench-MM.
title OmniSafeBench-MM: A Unified Benchmark and Toolbox for Multimodal Jailbreak Attack-Defense Evaluation
topic Cryptography and Security
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2512.06589