ASTRA: An Automated Framework for Strategy Discovery, Retrieval, and Evolution for Jailbreaking LLMs
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2025
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866917421611220992 |
|---|---|
| author | Liu, Xu Chen, Yan Ling, Kan Zhu, Yichi Zhang, Hengrun Fan, Guisheng Yu, Huiqun |
| author_facet | Liu, Xu Chen, Yan Ling, Kan Zhu, Yichi Zhang, Hengrun Fan, Guisheng Yu, Huiqun |
| contents | Despite extensive safety alignment, Large Language Models (LLMs) remain vulnerable to jailbreak attacks. However, existing methods generally lack the capability for continuous learning and self-evolution from interactions, limiting the diversity and adaptability of attack strategies. To address this, we propose ASTRA, an automated framework capable of autonomously discovering, retrieving, and evolving attack strategies. ASTRA operates on a closed-loop ``attack-evaluate-distill-reuse'' mechanism, which not only generates attack prompts but also automatically distills reusable strategies from every interaction. To systematically manage these strategies, we introduce a dynamic three-tier strategy library (Effective, Promising, and Ineffective) that categorizes strategies based on performance. This hierarchical memory mechanism enables the framework to enhance efficiency by leveraging successful patterns while optimizing the exploration space by avoiding known failures. Extensive experiments in a black-box setting demonstrate that ASTRA significantly outperforms existing baselines. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2511_02356 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | ASTRA: An Automated Framework for Strategy Discovery, Retrieval, and Evolution for Jailbreaking LLMs Liu, Xu Chen, Yan Ling, Kan Zhu, Yichi Zhang, Hengrun Fan, Guisheng Yu, Huiqun Cryptography and Security Machine Learning Despite extensive safety alignment, Large Language Models (LLMs) remain vulnerable to jailbreak attacks. However, existing methods generally lack the capability for continuous learning and self-evolution from interactions, limiting the diversity and adaptability of attack strategies. To address this, we propose ASTRA, an automated framework capable of autonomously discovering, retrieving, and evolving attack strategies. ASTRA operates on a closed-loop ``attack-evaluate-distill-reuse'' mechanism, which not only generates attack prompts but also automatically distills reusable strategies from every interaction. To systematically manage these strategies, we introduce a dynamic three-tier strategy library (Effective, Promising, and Ineffective) that categorizes strategies based on performance. This hierarchical memory mechanism enables the framework to enhance efficiency by leveraging successful patterns while optimizing the exploration space by avoiding known failures. Extensive experiments in a black-box setting demonstrate that ASTRA significantly outperforms existing baselines. |
| title | ASTRA: An Automated Framework for Strategy Discovery, Retrieval, and Evolution for Jailbreaking LLMs |
| topic | Cryptography and Security Machine Learning |
| url | https://arxiv.org/abs/2511.02356 |