Making Them Ask and Answer: Jailbreaking Large Language Models in Few Queries via Disguise and Reconstruction

TL;DR

This paper introduces DRA, a black-box attack method that exploits bias vulnerabilities in large language models to generate harmful responses, achieving high success rates across multiple models including GPT-4.

Contribution

It presents a novel theoretical framework and a practical attack method for jailbreaking LLMs by disguising and reconstructing harmful instructions.

Findings

DRA achieves a 91.1% success rate on GPT-4.

Effective across open-source and closed-source models.

State-of-the-art attack efficiency demonstrated.

Abstract

In recent years, large language models (LLMs) have demonstrated notable success across various tasks, but the trustworthiness of LLMs is still an open problem. One specific threat is the potential to generate toxic or harmful responses. Attackers can craft adversarial prompts that induce harmful responses from LLMs. In this work, we pioneer a theoretical foundation in LLMs security by identifying bias vulnerabilities within the safety fine-tuning and design a black-box jailbreak method named DRA (Disguise and Reconstruction Attack), which conceals harmful instructions through disguise and prompts the model to reconstruct the original harmful instruction within its completion. We evaluate DRA across various open-source and closed-source models, showcasing state-of-the-art jailbreak success rates and attack efficiency. Notably, DRA boasts a 91.1% attack success rate on OpenAI GPT-4 chatbot.