paper-with-me

Papers

Securing Vision-Language Models with a Robust Encoder Against Jailbreak and Adversarial Attacks

2024-09-11 · Md Zarif Hossain, Ahmed Imteaj

Large Vision-Language Models (LVLMs), trained on multimodal big datasets, have significantly advanced AI by excelling in vision-language tasks. However, these models remain vulnerable to adversarial attacks, particularly jailbreak attacks, which bypass safety protocols and cause the model to generate misleading or harmful responses. This vulnerability stems from both the inherent susceptibilities of LLMs and the expanded attack surface introduced by the visual modality. We propose Sim-CLIP+, a novel defense mechanism that adversarially fine-tunes the CLIP vision encoder by leveraging a Siamese architecture. This approach maximizes cosine similarity between perturbed and clean samples, facilitating resilience against adversarial manipulations. Sim-CLIP+ offers a plug-and-play solution, allowing seamless integration into existing LVLM architectures as a robust vision encoder. Unlike previous defenses, our method requires no structural modifications to the LVLM and incurs minimal computational overhead. Sim-CLIP+ demonstrates effectiveness against both gradient-based adversarial attacks and various jailbreak techniques. We evaluate Sim-CLIP+ against three distinct jailbreak attack strategies and perform clean evaluations using standard downstream datasets, including COCO for image captioning and OKVQA for visual question answering. Extensive experiments demonstrate that Sim-CLIP+ maintains high clean accuracy while substantially improving robustness against both gradient-based adversarial attacks and jailbreak techniques. Our code and robust vision encoders are available at https://github.com/speedlab-git/Robust-Encoder-against-Jailbreak-attack.git.

📄 PDF Abstract BibTeX arXiv:2409.07353

Code (0)

등록된 구현이 없습니다.

Tasks

Image CaptioningQuestion AnsweringVisual Question Answering

Methods 이 논문이 사용한 방법론

CLIP Contrastive Language-Image Pre-training (CLIP), consisting of a simplified version of ConVIRT trained from scratch, is an efficient method of image representation learning…

Similar Papers 제목 키워드 기반

Directional Embedding Smoothing for Robust Vision Language Models

2026-03-16 · Ye Wang, Jing Liu, Toshiaki Koike-Akino arxiv

The safety and reliability of vision-language models (VLMs) are a crucial part of deploying trustworthy agentic AI systems. However, VLMs remain vulnerable to jailbreaking attacks that undermine their safety alignment to…

JailBreakV: A Benchmark for Assessing the Robustness of MultiModal Large Language Models against Jailbreak Attacks

2024-04-03 · Weidi Luo, Siyuan Ma, Xiaogeng Liu, XIAOYU GUO 외

With the rapid advancements in Multimodal Large Language Models (MLLMs), securing these models against malicious inputs while aligning them with human values has emerged as a critical challenge. In this paper, we investi…

LLM Jailbreak

SafeAligner: Safety Alignment against Jailbreak Attacks via Response Disparity Guidance

2024-06-26 · Caishuang Huang, Wanxu Zhao, Rui Zheng, Huijie Lv 외

As the development of large language models (LLMs) rapidly advances, securing these models effectively without compromising their utility has become a pivotal area of research. However, current defense strategies against…

Safety Alignment

Preventing Robotic Jailbreaking via Multimodal Domain Adaptation

2025-09-27 · Francesco Marchiori, Rohan Sinha, Christopher Agia, Alexander Robey 외 arxiv

Large Language Models (LLMs) and Vision-Language Models (VLMs) are increasingly deployed in robotic environments but remain vulnerable to jailbreaking attacks that bypass safety mechanisms and drive unsafe or physically …

Autonomous DrivingDomain Adaptation

BlueSuffix: Reinforced Blue Teaming for Vision-Language Models Against Jailbreak Attacks

2024-10-28 · Yunhan Zhao, Xiang Zheng, Lin Luo, Yige Li 외

Despite their superb multimodal capabilities, Vision-Language Models (VLMs) have been shown to be vulnerable to jailbreak attacks, which are inference-time attacks that induce the model to output harmful responses with t…

2k