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 2408.11182cs.CRcs.AI
收藏

将恶意目标隐藏于良性叙述中:通过载体文章越狱大语言模型

Hide Your Malicious Goal Into Benign Narratives: Jailbreak Large Language Models through Carrier Articles

The Pennsylvania State University(宾夕法尼亚州立大学);Northern Arizona University(北亚利桑那大学);Worcester Polytechnic Institute(伍斯特理工学院);Carnegie Mellon University(卡内基梅隆大学)

Zhilong Wang, Haizhou Wang, Nanqing Luo, Lan Zhang, Xiaoyan Sun, Yebo Cao, Peng Liu

AI总结:

本文提出一种黑盒越狱方法,通过将禁止查询注入由上位词文章和上下文自动生成的良性载体文章来构造载荷提示,以激活相关语义神经元并抑制不良文本触发,在JailbreakBench上对四个模型测试平均成功率达63%,显著优于现有方法。

摘要:

大语言模型(LLM)越狱是指一类旨在绕过LLM安全防护、生成不符合安全使用指南内容的攻击。基于对自注意力计算过程的洞察,本文提出一种新颖的黑盒越狱方法,该方法通过将禁止查询策略性地注入载体文章来构造载荷提示。载体文章与禁止查询保持语义邻近性,其自动生成方式是结合由禁止查询生成的上位词文章和上下文。使用载体文章的直觉在于激活模型中与禁止查询语义相关的神经元,同时抑制会触发不良文本的神经元。载体文章本身是良性的,我们利用提示注入技术生成载荷提示。我们使用JailbreakBench评估所提方法,在100个不同的越狱目标上针对四个目标模型进行测试。实验结果表明,我们的方法具有优越的有效性,在所有目标模型上平均成功率达到63%,显著优于现有黑盒越狱方法。

英文摘要:

Large Language Model (LLM) jailbreak refers to a type of attack aimed to bypass the safeguard of an LLM to generate contents that are inconsistent with the safe usage guidelines. Based on the insights from the self-attention computation process, this paper proposes a novel blackbox jailbreak approach, which involves crafting the payload prompt by strategically injecting the prohibited query into a carrier article. The carrier article maintains the semantic proximity to the prohibited query, which is automatically produced by combining a hypernymy article and a context, both of which are generated from the prohibited query. The intuition behind the usage of carrier article is to activate the neurons in the model related to the semantics of the prohibited query while suppressing the neurons that will trigger the objectionable text. Carrier article itself is benign, and we leveraged prompt injection techniques to produce the payload prompt. We evaluate our approach using JailbreakBench, testing against four target models across 100 distinct jailbreak objectives. The experimental results demonstrate our method's superior effectiveness, achieving an average success rate of 63% across all target models, significantly outperforming existing blackbox jailbreak methods.

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。

↑