arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2410.02832cs.CRcs.AI
收藏

FlipAttack:通过翻转对大型语言模型进行劫持

FlipAttack: Jailbreak LLMs via Flipping

Engineering Programme, NUS Graduate School, National University of Singapore(国立新加坡大学整合科学与工程计划);Institute of Data Science (IDS), National University of Singapore(国立新加坡大学数据科学研究所);Department of Computer Science, School of Computing, National University of Singapore(国立新加坡大学计算机科学系)

Yue Liu, Xiaoxin He, Miao Xiong, Jinlan Fu, Shumin Deng, Yingwei Ma, Jiaheng Zhang, Bryan Hooi

AI总结:

FlipAttack通过在提示左侧添加噪声,利用语言模型自左向右理解文本的特性,实现对黑盒LLM的高效劫持,实验显示其在多个模型上均表现出高成功率。

摘要:

本文提出了一种简单而有效的针对黑盒大语言模型的劫持攻击方法FlipAttack。首先,基于自回归特性,我们发现语言模型倾向于从左到右理解文本,并发现当在左侧添加噪声时,它们难以理解文本。受这些见解启发,我们提出通过仅基于提示本身构造左侧噪声来伪装有害提示,然后将这一想法推广到4种翻转模式。其次,我们验证了语言模型执行文本翻转任务的强大能力,并开发了4种变体来指导语言模型去噪、理解和执行有害行为。这些设计使FlipAttack具有通用性、隐蔽性和简洁性,使其能够在仅1次查询中劫持黑盒大语言模型。在8个大语言模型上的实验展示了FlipAttack的优越性。值得注意的是,它在GPT-4o上实现了约98%的攻击成功率,并在平均情况下对5个防护模型实现了约98%的绕过率。代码可在GitHub上获取。

英文摘要:

This paper proposes a simple yet effective jailbreak attack named FlipAttack against black-box LLMs. First, from the autoregressive nature, we reveal that LLMs tend to understand the text from left to right and find that they struggle to comprehend the text when noise is added to the left side. Motivated by these insights, we propose to disguise the harmful prompt by constructing left-side noise merely based on the prompt itself, then generalize this idea to 4 flipping modes. Second, we verify the strong ability of LLMs to perform the text-flipping task, and then develop 4 variants to guide LLMs to denoise, understand, and execute harmful behaviors accurately. These designs keep FlipAttack universal, stealthy, and simple, allowing it to jailbreak black-box LLMs within only 1 query. Experiments on 8 LLMs demonstrate the superiority of FlipAttack. Remarkably, it achieves $\sim$98\% attack success rate on GPT-4o, and $\sim$98\% bypass rate against 5 guardrail models on average. The codes are available at GitHub\footnote{https://github.com/yueliu1999/FlipAttack}.

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。

↑