Layer-Level Self-Exposure and Patch: Affirmative Token Mitigation for Jailbreak Attack Defense

TL;DR

This paper presents Layer-AdvPatcher, a novel defense method that unlearns specific vulnerable layers in large language models to reduce jailbreak attack success rates while maintaining response utility.

Contribution

It introduces a self-augmented unlearning strategy targeting layer vulnerabilities to mitigate jailbreak attacks in LLMs, a novel approach in safety defense.

Findings

Reduces jailbreak success rate significantly

Maintains model utility on benign queries

Effective across multiple models and datasets

Abstract

As large language models (LLMs) are increasingly deployed in diverse applications, including chatbot assistants and code generation, aligning their behavior with safety and ethical standards has become paramount. However, jailbreak attacks, which exploit vulnerabilities to elicit unintended or harmful outputs, threaten LLMs' safety significantly. In this paper, we introduce Layer-AdvPatcher, a novel methodology designed to defend against jailbreak attacks by utilizing an unlearning strategy to patch specific layers within LLMs through self-augmented datasets. Our insight is that certain layer(s), tend to produce affirmative tokens when faced with harmful prompts. By identifying these layers and adversarially exposing them to generate more harmful data, one can understand their inherent and diverse vulnerabilities to attacks. With these exposures, we then "unlearn" these issues, reducing the impact of affirmative tokens and hence minimizing jailbreak risks while keeping the model's responses to safe queries intact. We conduct extensive experiments on two models, four benchmark datasets, and multiple state-of-the-art jailbreak attacks to demonstrate the efficacy of our approach. Results indicate that our framework reduces the harmfulness and attack success rate of jailbreak attacks without compromising utility for benign queries compared to recent defense methods. Our code is publicly available at: https://github.com/oyy2000/LayerAdvPatcher