Plentiful Jailbreaks with String Compositions

TL;DR

This paper introduces a framework of invertible string transformations to generate diverse and effective jailbreak attacks on large language models, revealing persistent vulnerabilities in current models.

Contribution

It unifies encoding-based attacks into a compositional framework and develops an automated method to generate numerous effective jailbreak strings.

Findings

High success rates on frontier models using the proposed attack.

Encoding-based attacks remain a significant vulnerability.

Automated composition sampling enhances attack diversity.

Abstract

Large language models (LLMs) remain vulnerable to a slew of adversarial attacks and jailbreaking methods. One common approach employed by white-hat attackers, or red-teamers, is to process model inputs and outputs using string-level obfuscations, which can include leetspeak, rotary ciphers, Base64, ASCII, and more. Our work extends these encoding-based attacks by unifying them in a framework of invertible string transformations. With invertibility, we can devise arbitrary string compositions, defined as sequences of transformations, that we can encode and decode end-to-end programmatically. We devise a automated best-of-n attack that samples from a combinatorially large number of string compositions. Our jailbreaks obtain competitive attack success rates on several leading frontier models when evaluated on HarmBench, highlighting that encoding-based attacks remain a persistent vulnerability even in advanced LLMs.