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 2512.04841cs.CRcs.AI
收藏

SoK:面向大语言模型安全的综合因果分析框架

SoK: a Comprehensive Causality Analysis Framework for Large Language Model Security

Singapore Management University(新加坡管理大学)

Wei Zhao, Zhe Li, Jun Sun

AI总结:

本文提出一个综合因果分析框架,用于研究大语言模型安全问题,通过系统分析因果因素提升模型安全性。

摘要:

大语言模型(LLMs)展现出卓越的能力,但仍然容易受到对抗性操纵的影响,如jailbreaking,其中精心设计的提示可以绕过安全机制。理解这些漏洞背后因果因素对于构建可靠的防御至关重要。在本工作中,我们引入了一个统一的因果分析框架,系统支持LLM中所有层次的因果研究,从token级、神经元级、层级干预到表示级分析。该框架使在多样化的因果攻击和防御方法之间进行一致的实验和比较成为可能。伴随这一实现,我们提供了首个关于因果驱动jailbreak研究的全面综述,并在多个开源模型和安全关键基准上经验性评估了该框架。我们的结果揭示了:(1)对因果关键组件的针对性干预可以可靠地改变安全行为;(2)与安全相关的机制高度局部化(即集中在早期到中期层,只有1-2%的神经元表现出因果影响);(3)从我们框架中提取的因果特征在多种威胁类型中实现了超过95%的检测准确率。通过将理论上的因果分析与实际模型安全相结合,我们的框架为研究基于因果的攻击、可解释性以及鲁棒攻击检测和缓解在LLM中的研究建立了可重复的基础。代码可在https://github.com/Amadeuszhao/SOK_Casuality获取。

英文摘要:

Large Language Models (LLMs) exhibit remarkable capabilities but remain vulnerable to adversarial manipulations such as jailbreaking, where crafted prompts bypass safety mechanisms. Understanding the causal factors behind such vulnerabilities is essential for building reliable defenses. In this work, we introduce a unified causality analysis framework that systematically supports all levels of causal investigation in LLMs, ranging from token-level, neuron-level, and layer-level interventions to representation-level analysis. The framework enables consistent experimentation and comparison across diverse causality-based attack and defense methods. Accompanying this implementation, we provide the first comprehensive survey of causality-driven jailbreak studies and empirically evaluate the framework on multiple open-weight models and safety-critical benchmarks including jailbreaks, hallucination detection, backdoor identification, and fairness evaluation. Our results reveal that: (1) targeted interventions on causally critical components can reliably modify safety behavior; (2) safety-related mechanisms are highly localized (i.e., concentrated in early-to-middle layers with only 1--2\% of neurons exhibiting causal influence); and (3) causal features extracted from our framework achieve over 95\% detection accuracy across multiple threat types. By bridging theoretical causality analysis and practical model safety, our framework establishes a reproducible foundation for research on causality-based attacks, interpretability, and robust attack detection and mitigation in LLMs. Code is available at https://github.com/Amadeuszhao/SOK_Casuality.

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。

↑