Align Is Not Enough: Multimodal Universal Jailbreak Attack Against Multimodal Large Language Models
- 6 January 2025
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems for Video Technology
- Vol. 35 (6) , 5475-5488
Abstract
Large Language Models (LLMs) have evolved into Multimodal Large Language Models (MLLMs), significantly enhancing their capabilities by integrating visual information and other types, thus aligning more closely with the nature of human intelligence, which processes a variety of data forms beyond just text. Despite advancements, the undesirable generation of these models remains a critical concern, particularly due to vulnerabilities exposed by text-based jailbreak attacks, which have represented a significant threat by challenging existing safety protocols. Motivated by the unique security risks posed by the integration of new and old modalities for MLLMs, we propose a unified multimodal universal jailbreak attack framework that leverages iterative image-text interactions and transfer-based strategy to generate a universal adversarial suffix and image. Our work not only highlights the interaction of image-text modalities can be used as a critical vulnerability but also validates that multimodal universal jailbreak attacks can bring higher-quality undesirable generations across different MLLMs. We evaluate the undesirable context generation of MLLMs like LLaVA, Yi-VL, MiniGPT4, MiniGPT-v2, and InstructBLIP, and reveal significant multimodal safety alignment issues, highlighting the inadequacy of current safety mechanisms against sophisticated multimodal attacks. This study underscores the urgent need for robust safety measures in MLLMs, advocating for a comprehensive review and enhancement of security protocols to mitigate potential risks associated with multimodal capabilities.Keywords
All Related Versions
Funding Information
- National Natural Science Foundation of China (62306098)
- National Natural Science Foundation of China (U23B2031)
- Open Projects Program of State Key Laboratory of Multimodal Artificial Intelligence Systems
- Fundamental Research Funds for the Central Universities (JZ2024HGTB0256)
- National Natural Science Foundation, Singapore under its AI Singapore Programme (AISG2-RP-2021-022)
This publication has 13 references indexed in Scilit:
- A survey on multimodal large language modelsNational Science Review, 2024
- Scalable Universal Adversarial Watermark Defending Against Facial ForgeryIEEE Transactions on Information Forensics and Security, 2024
- mPLUG-OwI2: Revolutionizing Multi-modal Large Language Model with Modality Collaboration2024
- Dynamics-Aware Adversarial Attack of Adaptive Neural NetworksIEEE Transactions on Circuits and Systems for Video Technology, 2024
- Iterative Adversarial Attack on Image-Guided Story Ending GenerationIEEE Transactions on Multimedia, 2023
- Cross-Shaped Adversarial Patch AttackIEEE Transactions on Circuits and Systems for Video Technology, 2023
- Toward Robust Neural Image Compression: Adversarial Attack and Model FinetuningIEEE Transactions on Circuits and Systems for Video Technology, 2023
- Targeted Adversarial Attack Against Deep Cross-Modal Hashing RetrievalIEEE Transactions on Circuits and Systems for Video Technology, 2023
- A novel model watermarking for protecting generative adversarial networkComputers & Security, 2023
- Enhancing the Transferability of Adversarial Attacks through Variance Tuning2021