MIDAS: Multi-Image Dispersion and Semantic Reconstruction for Jailbreaking MLLMs

TL;DR

The paper introduces MIDAS, a multimodal jailbreak framework that disperses harmful semantics across multiple images and uses cross-image reasoning to bypass safety mechanisms in MLLMs, significantly increasing attack success rates.

Contribution

MIDAS is the first framework to decompose and disperse malicious content across multiple images, enabling more effective jailbreak attacks on advanced MLLMs.

Findings

Achieves an average attack success rate of 81.46% across 4 closed-source MLLMs.

Outperforms state-of-the-art jailbreak methods in effectiveness.

Enforces longer, structured multi-image reasoning to bypass safety defenses.

Abstract

Multimodal Large Language Models (MLLMs) have achieved remarkable performance but remain vulnerable to jailbreak attacks that can induce harmful content and undermine their secure deployment. Previous studies have shown that introducing additional inference steps, which disrupt security attention, can make MLLMs more susceptible to being misled into generating malicious content. However, these methods rely on single-image masking or isolated visual cues, which only modestly extend reasoning paths and thus achieve limited effectiveness, particularly against strongly aligned commercial closed-source models. To address this problem, in this paper, we propose Multi-Image Dispersion and Semantic Reconstruction (MIDAS), a multimodal jailbreak framework that decomposes harmful semantics into risk-bearing subunits, disperses them across multiple visual clues, and leverages cross-image reasoning to gradually reconstruct the malicious intent, thereby bypassing existing safety mechanisms. The proposed MIDAS enforces longer and more structured multi-image chained reasoning, substantially increases the model's reliance on visual cues while delaying the exposure of malicious semantics and significantly reducing the model's security attention, thereby improving the performance of jailbreak against advanced MLLMs. Extensive experiments across different datasets and MLLMs demonstrate that the proposed MIDAS outperforms state-of-the-art jailbreak attacks for MLLMs and achieves an average attack success rate of 81.46% across 4 closed-source MLLMs. Our code is available at this [link](https://github.com/Winnie-Lian/MIDAS).