arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2512.03356cs.CR
收藏

从静态到适应性:基于免疫记忆的大型语言模型 jailbreak 检测

From static to adaptive: immune memory-based jailbreak detection for large language models

Beijing University of Posts and Telecommunications(北京邮电大学);Hunan Branch of National Computer Network Emergency Response(国家计算机网络应急技术处理协调中心湖南分中心);Clinical Oncology School of Fujian Medical University(福建医科大学临床肿瘤学院)

Jun Leng, Yu Liu, Litian Zhang, Ruihan Hu, Zhuting Fang, Xi Zhang

AI总结:

基于免疫记忆的框架,通过动态记忆更新和主动免疫机制,提升大型语言模型对 jailbreak 攻击的检测准确率至 94%。

摘要:

大型语言模型(LLMs)是现代人工智能系统的核心,但它们仍然容易受到对抗性 jailbreak 攻击。因此,对这类恶意输入进行稳健检测对于确保模型安全至关重要。传统检测方法通常依赖于在固定大规模数据集上训练的外部模型,这通常会带来显著的计算开销。尽管最近的方法转向利用模型内部的安全信号以实现更轻量和高效的检测,但这些方法本质上仍是静态的,难以适应 jailbreak 攻击不断演变的性质。受生物免疫机制的启发,我们引入了免疫记忆适应守护(IMAG)框架。通过提炼和编码安全模式到一个持久且可进化记忆库中,IMAG 能够适应新兴威胁。具体而言,该框架协调三个协同作用的组件:免疫检测,利用检索实现对已知 jailbreak 攻击的高效拦截;主动免疫,通过主动行为模拟解决模糊的未知查询;记忆更新,将验证过的攻击模式重新整合到记忆库中。这种闭环架构将 LLM 的防御从刚性过滤转变为自主适应性缓解。在五个代表性开源 LLM 上的广泛评估表明,我们的方法超越了最先进的(SOTA)基线,实现了在多样且复杂的攻击类型上平均检测准确率高达 94% 的优异表现。

英文摘要:

Large Language Models (LLMs) serve as the backbone of modern AI systems, yet they remain susceptible to adversarial jailbreak attacks. Consequently, robust detection of such malicious inputs is paramount for ensuring model safety. Traditional detection methods typically rely on external models trained on fixed, large-scale datasets, which often incur significant computational overhead. While recent methods shift toward leveraging internal safety signals of models to enable more lightweight and efficient detection. However, these methods remain inherently static and struggle to adapt to the evolving nature of jailbreak attacks. Drawing inspiration from the biological immune mechanism, we introduce the Immune Memory Adaptive Guard (IMAG) framework. By distilling and encoding safety patterns into a persistent, evolvable memory bank, IMAG enables adaptive generalization to emerging threats. Specifically, the framework orchestrates three synergistic components: Immune Detection, which employs retrieval for high-efficiency interception of known jailbreak attacks; Active Immunity, which performs proactive behavioral simulation to resolve ambiguous unknown queries; Memory Updating, which integrates validated attack patterns back into the memory bank. This closed-loop architecture transitions LLM defense from rigid filtering to autonomous adaptive mitigation. Extensive evaluations across five representative open-source LLMs demonstrate that our method surpasses state-of-the-art (SOTA) baselines, achieving a superior average detection accuracy of 94\% across diverse and complex attack types.

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。

↑