CRAICLLGOct 23, 2023

AutoDAN: Interpretable Gradient-Based Adversarial Attacks on Large Language Models

arXiv:2310.15140v2109 citationsh-index: 52
Originality Highly original
AI Analysis

This work addresses the vulnerability of LLMs to adversarial attacks, offering a new method for red-teaming and understanding jailbreak mechanisms, though it is incremental as it builds on existing attack types.

The paper tackles the problem of compromising safety alignment in Large Language Models (LLMs) by introducing AutoDAN, an interpretable gradient-based adversarial attack that generates readable prompts, achieving high attack success rates while bypassing perplexity filters and generalizing to unforeseen harmful behaviors and black-box LLMs.

Safety alignment of Large Language Models (LLMs) can be compromised with manual jailbreak attacks and (automatic) adversarial attacks. Recent studies suggest that defending against these attacks is possible: adversarial attacks generate unlimited but unreadable gibberish prompts, detectable by perplexity-based filters; manual jailbreak attacks craft readable prompts, but their limited number due to the necessity of human creativity allows for easy blocking. In this paper, we show that these solutions may be too optimistic. We introduce AutoDAN, an interpretable, gradient-based adversarial attack that merges the strengths of both attack types. Guided by the dual goals of jailbreak and readability, AutoDAN optimizes and generates tokens one by one from left to right, resulting in readable prompts that bypass perplexity filters while maintaining high attack success rates. Notably, these prompts, generated from scratch using gradients, are interpretable and diverse, with emerging strategies commonly seen in manual jailbreak attacks. They also generalize to unforeseen harmful behaviors and transfer to black-box LLMs better than their unreadable counterparts when using limited training data or a single proxy model. Furthermore, we show the versatility of AutoDAN by automatically leaking system prompts using a customized objective. Our work offers a new way to red-team LLMs and understand jailbreak mechanisms via interpretability.

Code Implementations1 repo
Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes