使用多片段视频破解多模态大语言模型
Jailbreaking Multimodal Large Language Models using Multi-Clip Video
AI总结：
提出MCV SafetyBench数据集，通过多片段视频评估多模态大语言模型的安全漏洞，发现视频模态比图像更脆弱，动态和多样化上下文增加攻击成功率，并基于图像模态的鲁棒性提出防御策略。
摘要：
随着多模态大语言模型（MLLMs）发展到处理视频输入，人们开始担忧其被恶意滥用的可能性。先前的越狱研究表明，MLLMs中的安全对齐可以通过视觉输入被绕过，但尚不清楚视频输入的哪些属性导致了这种脆弱性。为填补这一空白，我们引入了Multi-Clip Video (MCV) SafetyBench，一个包含2,920个视频的数据集，旨在评估视频输入的多样性如何影响MLLMs的脆弱性。每个视频由多个短片段组成，描述与有害查询相关的不同上下文。对八个代表性视频MLLMs的实验表明，攻击成功率随着片段数量的增加而持续提高。我们的结果进一步表明，视频模态（1）比图像模态更脆弱，（2）对动态视频比对静态视频更脆弱，（3）当视频包含更多样化的上下文时更脆弱。基于这些发现，我们提出了一种利用图像模态相对鲁棒性的防御策略。
英文摘要：
As multimodal large language models (MLLMs) have advanced to process video inputs, concerns have emerged about their potential for malicious misuse. Prior jailbreak studies have shown that safety alignment in MLLMs can be bypassed through visual inputs, yet it remains unclear which properties of video inputs induce this vulnerability. To address this gap, we introduce Multi-Clip Video (MCV) SafetyBench, a dataset of 2,920 videos designed to evaluate how the diversity of video inputs affects the vulnerability of MLLMs. Each video consists of multiple short clips depicting diverse contexts related to a harmful query. Experiments on eight representative video MLLMs show that attack success consistently increases with the number of clips. Our results further indicate that the video modality is (1) more vulnerable than the image modality, (2) more vulnerable to dynamic videos than to static videos, and (3) more vulnerable when videos contain more diverse contexts. Building on these findings, we propose a defense strategy that leverages the relative robustness of the image modality.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
