arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2404.04849cs.CRcs.AI
收藏

将你的恶意目标隐藏在良性叙述中:通过逻辑链注入破解大语言模型

Hidden You Malicious Goal Into Benign Narratives: Jailbreak Large Language Models through Logic Chain Injection

The Pennsylvania State University(宾夕法尼亚州立大学);Carnegie Mellon University(卡内基梅隆大学)

Zhilong Wang, Yebo Cao, Peng Liu

AI总结:

本研究提出一种逻辑链注入式越狱攻击,借鉴社会心理学“谎言藏于真相更易骗人”的原理,将恶意目标拆解为良性叙述链嵌入含事实的良性文章,可同时欺骗大语言模型与人类安全分析师,为LLM安全防护提出新挑战。

摘要:

针对大语言模型(LLM)的越狱攻击指的是设计旨在利用模型生成恶意内容的提示词。现有越狱攻击虽能成功欺骗大语言模型,但无法骗过人类。本文提出一种新型越狱攻击,既能欺骗大语言模型,也能骗过人类(即安全分析师)。该思路的核心洞见借鉴自社会心理学——即如果谎言藏在真相中,人类很容易受骗。基于这一洞见,我们提出逻辑链注入攻击,将恶意意图注入到良性真相中。逻辑链注入攻击首先将其恶意目标拆解为一条良性叙述链,随后将这些叙述分散到一篇包含确凿事实的相关良性文章中。通过这种方式,新生成的提示词不仅能欺骗大语言模型,还能骗过人类。

英文摘要:

Jailbreak attacks on Language Model Models (LLMs) entail crafting prompts aimed at exploiting the models to generate malicious content. Existing jailbreak attacks can successfully deceive the LLMs, however they cannot deceive the human. This paper proposes a new type of jailbreak attacks which can deceive both the LLMs and human (i.e., security analyst). The key insight of our idea is borrowed from the social psychology - that is human are easily deceived if the lie is hidden in truth. Based on this insight, we proposed the logic-chain injection attacks to inject malicious intention into benign truth. Logic-chain injection attack firstly dissembles its malicious target into a chain of benign narrations, and then distribute narrations into a related benign article, with undoubted facts. In this way, newly generate prompt cannot only deceive the LLMs, but also deceive human.

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。

↑