PRISM: Programmatic Reasoning with Image Sequence Manipulation for LVLM Jailbreaking

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zou, Quanchen, Ying, Zonghao, Chen, Moyang, Xu, Wenzhuo, Xiao, Yisong, Li, Yakai, Zhang, Deyue, Yang, Dongdong, Liu, Zhao, Zhang, Xiangzheng
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914456745803776
author Zou, Quanchen
Ying, Zonghao
Chen, Moyang
Xu, Wenzhuo
Xiao, Yisong
Li, Yakai
Zhang, Deyue
Yang, Dongdong
Liu, Zhao
Zhang, Xiangzheng
author_facet Zou, Quanchen
Ying, Zonghao
Chen, Moyang
Xu, Wenzhuo
Xiao, Yisong
Li, Yakai
Zhang, Deyue
Yang, Dongdong
Liu, Zhao
Zhang, Xiangzheng
contents The increasing sophistication of large vision-language models (LVLMs) has been accompanied by advances in safety alignment mechanisms designed to prevent harmful content generation. However, these defenses remain vulnerable to sophisticated adversarial attacks. Existing jailbreak methods typically rely on direct and semantically explicit prompts, overlooking subtle vulnerabilities in how LVLMs compose information over multiple reasoning steps. In this paper, we propose a novel and effective jailbreak framework inspired by Return-Oriented Programming (ROP) techniques from software security. Our approach decomposes a harmful instruction into a sequence of individually benign visual gadgets. A carefully engineered textual prompt directs the sequence of inputs, prompting the model to integrate the benign visual gadgets through its reasoning process to produce a coherent and harmful output. This makes the malicious intent emergent and difficult to detect from any single component. We validate our method through extensive experiments on established benchmarks including SafeBench and MM-SafetyBench, targeting popular LVLMs. Results show that our approach consistently and substantially outperforms existing baselines on state-of-the-art models, achieving near-perfect attack success rates (over 0.90 on SafeBench) and improving ASR by up to 0.39. Our findings reveal a critical and underexplored vulnerability that exploits the compositional reasoning abilities of LVLMs, highlighting the urgent need for defenses that secure the entire reasoning process.
format Preprint
id arxiv_https___arxiv_org_abs_2507_21540
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle PRISM: Programmatic Reasoning with Image Sequence Manipulation for LVLM Jailbreaking
Zou, Quanchen
Ying, Zonghao
Chen, Moyang
Xu, Wenzhuo
Xiao, Yisong
Li, Yakai
Zhang, Deyue
Yang, Dongdong
Liu, Zhao
Zhang, Xiangzheng
Cryptography and Security
Computer Vision and Pattern Recognition
The increasing sophistication of large vision-language models (LVLMs) has been accompanied by advances in safety alignment mechanisms designed to prevent harmful content generation. However, these defenses remain vulnerable to sophisticated adversarial attacks. Existing jailbreak methods typically rely on direct and semantically explicit prompts, overlooking subtle vulnerabilities in how LVLMs compose information over multiple reasoning steps. In this paper, we propose a novel and effective jailbreak framework inspired by Return-Oriented Programming (ROP) techniques from software security. Our approach decomposes a harmful instruction into a sequence of individually benign visual gadgets. A carefully engineered textual prompt directs the sequence of inputs, prompting the model to integrate the benign visual gadgets through its reasoning process to produce a coherent and harmful output. This makes the malicious intent emergent and difficult to detect from any single component. We validate our method through extensive experiments on established benchmarks including SafeBench and MM-SafetyBench, targeting popular LVLMs. Results show that our approach consistently and substantially outperforms existing baselines on state-of-the-art models, achieving near-perfect attack success rates (over 0.90 on SafeBench) and improving ASR by up to 0.39. Our findings reveal a critical and underexplored vulnerability that exploits the compositional reasoning abilities of LVLMs, highlighting the urgent need for defenses that secure the entire reasoning process.
title PRISM: Programmatic Reasoning with Image Sequence Manipulation for LVLM Jailbreaking
topic Cryptography and Security
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2507.21540