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 2502.01154cs.CLcs.AIcs.CRcs.LG
收藏

使用通用多提示词进行越狱

Jailbreaking with Universal Multi-Prompts

National Taiwan University(台湾大学)

Yu-Ling Hsu, Hsuan Su, Shang-Tse Chen

AI总结:

针对大型语言模型的越狱攻击研究,提出基于通用多提示词的攻击方法JUMP及其防御适配DUMP,实验证明该方法优于现有技术。

摘要:

大型语言模型(LLMs)近年来发展迅速,彻底改变了各种应用,并显著提升了便利性和生产力。然而,在其令人印象深刻的能力之外,伦理问题和新类型的攻击(如越狱)也随之出现。虽然大多数提示技术侧重于为单个案例优化对抗性输入,但在处理大型数据集时会导致更高的计算成本。较少有研究探讨训练能够迁移到未见任务的通用攻击者这一更普遍的设置。在本文中,我们引入了JUMP,这是一种基于提示的方法,旨在使用通用多提示词对LLMs进行越狱。我们还调整了我们的方法用于防御,我们将其称为DUMP。实验结果表明,我们优化通用多提示词的方法优于现有技术。

英文摘要:

Large language models (LLMs) have seen rapid development in recent years, revolutionizing various applications and significantly enhancing convenience and productivity. However, alongside their impressive capabilities, ethical concerns and new types of attacks, such as jailbreaking, have emerged. While most prompting techniques focus on optimizing adversarial inputs for individual cases, resulting in higher computational costs when dealing with large datasets. Less research has addressed the more general setting of training a universal attacker that can transfer to unseen tasks. In this paper, we introduce JUMP, a prompt-based method designed to jailbreak LLMs using universal multi-prompts. We also adapt our approach for defense, which we term DUMP. Experimental results demonstrate that our method for optimizing universal multi-prompts outperforms existing techniques.

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。

↑