Activation-Guided Local Editing for Jailbreaking Attacks

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wang, Jiecong, Li, Haoran, Peng, Hao, Zeng, Ziqian, Wang, Zihao, Du, Haohua, Yu, Zhengtao
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911593319628800
author Wang, Jiecong
Li, Haoran
Peng, Hao
Zeng, Ziqian
Wang, Zihao
Du, Haohua
Yu, Zhengtao
author_facet Wang, Jiecong
Li, Haoran
Peng, Hao
Zeng, Ziqian
Wang, Zihao
Du, Haohua
Yu, Zhengtao
contents Jailbreaking is an essential adversarial technique for red-teaming these models to uncover and patch security flaws. However, existing jailbreak methods face significant drawbacks. Token-level jailbreak attacks often produce incoherent or unreadable inputs and exhibit poor transferability, while prompt-level attacks lack scalability and rely heavily on manual effort and human ingenuity. We propose a concise and effective two-stage framework that combines the advantages of these approaches. The first stage performs a scenario-based generation of context and rephrases the original malicious query to obscure its harmful intent. The second stage then utilizes information from the model's hidden states to guide fine-grained edits, effectively steering the model's internal representation of the input from a malicious toward a benign one. Extensive experiments demonstrate that this method achieves state-of-the-art Attack Success Rate, with gains of up to 37.74% over the strongest baseline, and exhibits excellent transferability to black-box models. Our analysis further demonstrates that AGILE maintains substantial effectiveness against prominent defense mechanisms, highlighting the limitations of current safeguards and providing valuable insights for future defense development. Our code is available at https://github.com/SELGroup/AGILE.
format Preprint
id arxiv_https___arxiv_org_abs_2508_00555
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Activation-Guided Local Editing for Jailbreaking Attacks
Wang, Jiecong
Li, Haoran
Peng, Hao
Zeng, Ziqian
Wang, Zihao
Du, Haohua
Yu, Zhengtao
Cryptography and Security
Artificial Intelligence
Computation and Language
Jailbreaking is an essential adversarial technique for red-teaming these models to uncover and patch security flaws. However, existing jailbreak methods face significant drawbacks. Token-level jailbreak attacks often produce incoherent or unreadable inputs and exhibit poor transferability, while prompt-level attacks lack scalability and rely heavily on manual effort and human ingenuity. We propose a concise and effective two-stage framework that combines the advantages of these approaches. The first stage performs a scenario-based generation of context and rephrases the original malicious query to obscure its harmful intent. The second stage then utilizes information from the model's hidden states to guide fine-grained edits, effectively steering the model's internal representation of the input from a malicious toward a benign one. Extensive experiments demonstrate that this method achieves state-of-the-art Attack Success Rate, with gains of up to 37.74% over the strongest baseline, and exhibits excellent transferability to black-box models. Our analysis further demonstrates that AGILE maintains substantial effectiveness against prominent defense mechanisms, highlighting the limitations of current safeguards and providing valuable insights for future defense development. Our code is available at https://github.com/SELGroup/AGILE.
title Activation-Guided Local Editing for Jailbreaking Attacks
topic Cryptography and Security
Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2508.00555