Boosting Jailbreak Transferability for Large Language Models

TL;DR

This paper introduces a novel method to improve the transferability of jailbreak attacks on large language models, achieving near-perfect success rates and winning a global challenge.

Contribution

We propose enhancements like scenario induction, optimized suffix selection, and re-suffix mechanisms to significantly boost jailbreak transferability across models.

Findings

Achieved nearly 100% success in attack transferability.

Outperformed existing methods in extensive benchmarks.

Won first place in the AISG global challenge.

Abstract

Large language models have drawn significant attention to the challenge of safe alignment, especially regarding jailbreak attacks that circumvent security measures to produce harmful content. To address the limitations of existing methods like GCG, which perform well in single-model attacks but lack transferability, we propose several enhancements, including a scenario induction template, optimized suffix selection, and the integration of re-suffix attack mechanism to reduce inconsistent outputs. Our approach has shown superior performance in extensive experiments across various benchmarks, achieving nearly 100% success rates in both attack execution and transferability. Notably, our method has won the first place in the AISG-hosted Global Challenge for Safe and Secure LLMs. The code is released at https://github.com/HqingLiu/SI-GCG.