Saved in:
Bibliographic Details
Main Author: Huang, Brian R. Y.
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2411.01084
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917865633873920
author Huang, Brian R. Y.
author_facet Huang, Brian R. Y.
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2411_01084
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Plentiful Jailbreaks with String Compositions
Huang, Brian R. Y.
Computation and Language
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.
title Plentiful Jailbreaks with String Compositions
topic Computation and Language
url https://arxiv.org/abs/2411.01084