AutoJailbreak: Exploring Jailbreak Attacks and Defenses through a Dependency Lens
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910474802561024 |
|---|---|
| author | Lu, Lin Yan, Hai Yuan, Zenghui Shi, Jiawen Wei, Wenqi Chen, Pin-Yu Zhou, Pan |
| author_facet | Lu, Lin Yan, Hai Yuan, Zenghui Shi, Jiawen Wei, Wenqi Chen, Pin-Yu Zhou, Pan |
| contents | Jailbreak attacks in large language models (LLMs) entail inducing the models to generate content that breaches ethical and legal norm through the use of malicious prompts, posing a substantial threat to LLM security. Current strategies for jailbreak attack and defense often focus on optimizing locally within specific algorithmic frameworks, resulting in ineffective optimization and limited scalability. In this paper, we present a systematic analysis of the dependency relationships in jailbreak attack and defense techniques, generalizing them to all possible attack surfaces. We employ directed acyclic graphs (DAGs) to position and analyze existing jailbreak attacks, defenses, and evaluation methodologies, and propose three comprehensive, automated, and logical frameworks. \texttt{AutoAttack} investigates dependencies in two lines of jailbreak optimization strategies: genetic algorithm (GA)-based attacks and adversarial-generation-based attacks, respectively. We then introduce an ensemble jailbreak attack to exploit these dependencies. \texttt{AutoDefense} offers a mixture-of-defenders approach by leveraging the dependency relationships in pre-generative and post-generative defense strategies. \texttt{AutoEvaluation} introduces a novel evaluation method that distinguishes hallucinations, which are often overlooked, from jailbreak attack and defense responses. Through extensive experiments, we demonstrate that the proposed ensemble jailbreak attack and defense framework significantly outperforms existing research. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2406_03805 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | AutoJailbreak: Exploring Jailbreak Attacks and Defenses through a Dependency Lens Lu, Lin Yan, Hai Yuan, Zenghui Shi, Jiawen Wei, Wenqi Chen, Pin-Yu Zhou, Pan Cryptography and Security Jailbreak attacks in large language models (LLMs) entail inducing the models to generate content that breaches ethical and legal norm through the use of malicious prompts, posing a substantial threat to LLM security. Current strategies for jailbreak attack and defense often focus on optimizing locally within specific algorithmic frameworks, resulting in ineffective optimization and limited scalability. In this paper, we present a systematic analysis of the dependency relationships in jailbreak attack and defense techniques, generalizing them to all possible attack surfaces. We employ directed acyclic graphs (DAGs) to position and analyze existing jailbreak attacks, defenses, and evaluation methodologies, and propose three comprehensive, automated, and logical frameworks. \texttt{AutoAttack} investigates dependencies in two lines of jailbreak optimization strategies: genetic algorithm (GA)-based attacks and adversarial-generation-based attacks, respectively. We then introduce an ensemble jailbreak attack to exploit these dependencies. \texttt{AutoDefense} offers a mixture-of-defenders approach by leveraging the dependency relationships in pre-generative and post-generative defense strategies. \texttt{AutoEvaluation} introduces a novel evaluation method that distinguishes hallucinations, which are often overlooked, from jailbreak attack and defense responses. Through extensive experiments, we demonstrate that the proposed ensemble jailbreak attack and defense framework significantly outperforms existing research. |
| title | AutoJailbreak: Exploring Jailbreak Attacks and Defenses through a Dependency Lens |
| topic | Cryptography and Security |
| url | https://arxiv.org/abs/2406.03805 |