Metraipolis › archive › dc_23acb15e8554
Diagnose Before You Refine: Multi-Dimensional Judge Feedback for Query-Efficient Jailbreak Attacks — arXiv:2610.19971
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.19971
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-15T21:46:40.102Z
- SHA-256 of the archived text
e806de4da335b76cb4e52e9f6bc298a823d0b39496b1c76fded29b920f0ad702- Stable URL of this page
https://metraipolis.com/documento/dc_23acb15e8554
How to cite
Diagnose Before You Refine: Multi-Dimensional Judge Feedback for Query-Efficient Jailbreak Attacks — arXiv:2610.19971 — arXiv (en), https://arxiv.org/abs/2610.19971 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_23acb15e8554 (Metraipolis archive dc_23acb15e8554, sha256 e806de4da335b76cb4e52e9f6bc298a823d0b39496b1c76fded29b920f0ad702)
BibTeX
@misc{dc23acb15e8554,
title = {Diagnose Before You Refine: Multi-Dimensional Judge Feedback for Query-Efficient Jailbreak Attacks — arXiv:2610.19971},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.19971},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_23acb15e8554; sha256 e806de4da335b76cb4e52e9f6bc298a823d0b39496b1c76fded29b920f0ad702},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.19971. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Automated jailbreak attacks rely on evaluation feedback to optimize adversarial prompts, but existing methods often lack fine-grained diagnostic information. We propose Jev-Guided Strategy Bandit (JSB), a query-based jailbreak framework that leverages multi-dimensional probabilistic feedback to guide prompt optimization. JSB combines continuous jailbreak scoring with targeted refinement, strategy switching, and best-state reseeding to balance exploration and refinement. We evaluate JSB on two harmful request datasets across five open-source and three closed-source LLMs. Under a budget of only 10 target-model queries, JSB achieves an average HarmBench attack success rate of 89.8\% across ten open-source settings, consistently outperforming existing automated jailbreak attacks. Further analyses demonstrate its effectiveness under varying query budgets and attacker models. Defense experiments also reveal limitations in existing filtering mechanisms and highlight the potential of Jev-based output filtering. Tema degli esploratori: fiducia digitale e crittografia (elenco arXiv cs.CR, nuovi lavori del giorno). Categorie arXiv: Cryptography and Security (cs.CR). Pagina del lavoro: https://arxiv.org/abs/2610.19971 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-15T21:33:44.514Z dalla pagina https://arxiv.org/list/cs.CR/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).