Buchumschlag

Short-length Adversarial Training Helps LLMs Defend Long-length Jailbreak Attacks: Theoretical and Empirical Evidence

Fuente: arXiv

Gespeichert in:

Bibliographische Detailangaben
Hauptverfasser:	Fu, Shaopeng, Ding, Liang, Zhang, Jingfeng, Wang, Di
Format:	Preprint
Veröffentlicht:	2025
Schlagworte:	Machine Learning Cryptography and Security
Online-Zugang:	Acceder al recurso
Tags:	Tag hinzufügen Keine Tags, Fügen Sie den ersten Tag hinzu!

Ähnliche Einträge

Understanding and Improving Continuous Adversarial Training for LLMs via In-context Learning Theory
von: Fu, Shaopeng, et al.
Veröffentlicht: (2026)

Defending Jailbreak Prompts via In-Context Adversarial Game
von: Zhou, Yujun, et al.
Veröffentlicht: (2024)

Efficient Adversarial Training in LLMs with Continuous Attacks
von: Xhonneux, Sophie, et al.
Veröffentlicht: (2024)

From Theft to Bomb-Making: The Ripple Effect of Unlearning in Defending Against Jailbreak Attacks
von: Zhang, Zhexin, et al.
Veröffentlicht: (2024)

HSF: Defending against Jailbreak Attacks with Hidden State Filtering
von: Qian, Cheng, et al.
Veröffentlicht: (2024)

HashVFL: Defending Against Data Reconstruction Attacks in Vertical Federated Learning
von: Qiu, Pengyu, et al.
Veröffentlicht: (2022)

Low-Cost Hard-Label Adversarial Attack with Theoretical Foundations
von: Liu, Jun, et al.
Veröffentlicht: (2026)

DataFreeShield: Defending Adversarial Attacks without Training Data
von: Lee, Hyeyoon, et al.
Veröffentlicht: (2024)

MEA-Defender: A Robust Watermark against Model Extraction Attack
von: Lv, Peizhuo, et al.
Veröffentlicht: (2024)

Adversarial Tuning: Defending Against Jailbreak Attacks for LLMs
von: Liu, Fan, et al.
Veröffentlicht: (2024)

MIST: Defending Against Membership Inference Attacks Through Membership-Invariant Subspace Training
von: Li, Jiacheng, et al.
Veröffentlicht: (2023)

Invariant Aggregator for Defending against Federated Backdoor Attacks
von: Wang, Xiaoyang, et al.
Veröffentlicht: (2022)

Defending Against Unforeseen Failure Modes with Latent Adversarial Training
von: Casper, Stephen, et al.
Veröffentlicht: (2024)

Defending Against Sophisticated Poisoning Attacks with RL-based Aggregation in Federated Learning
von: Wang, Yujing, et al.
Veröffentlicht: (2024)

Generalization Properties of Adversarial Training for $\ell_0$-Bounded Adversarial Attacks
von: Delgosha, Payam, et al.
Veröffentlicht: (2024)

Pre-trained Encoder Inference: Revealing Upstream Encoders In Downstream Machine Learning Services
von: Fu, Shaopeng, et al.
Veröffentlicht: (2024)

LARGO: Latent Adversarial Reflection through Gradient Optimization for Jailbreaking LLMs
von: Li, Ran, et al.
Veröffentlicht: (2025)

FL-Defender: Combating Targeted Attacks in Federated Learning
von: Jebreel, Najeeb, et al.
Veröffentlicht: (2022)

Defending against Backdoor Attack on Deep Neural Networks
von: Cheng, Hao, et al.
Veröffentlicht: (2020)

How to Defend Against Large-scale Model Poisoning Attacks in Federated Learning: A Vertical Solution
von: Wang, Jinbo, et al.
Veröffentlicht: (2024)

Understanding and Enhancing the Transferability of Jailbreaking Attacks
von: Lin, Runqi, et al.
Veröffentlicht: (2025)

JailbreakRadar: Comprehensive Assessment of Jailbreak Attacks Against LLMs
von: Chu, Junjie, et al.
Veröffentlicht: (2024)

Enhancing Adversarial Attacks via Parameter Adaptive Adversarial Attack
von: Jin, Zhibo, et al.
Veröffentlicht: (2024)

Voice Jailbreak Attacks Against GPT-4o
von: Shen, Xinyue, et al.
Veröffentlicht: (2024)

MetaDefense: Defending Finetuning-based Jailbreak Attack Before and During Generation
von: Jiang, Weisen, et al.
Veröffentlicht: (2025)

Bits Leaked per Query: Information-Theoretic Bounds on Adversarial Attacks against LLMs
von: Kaneko, Masahiro, et al.
Veröffentlicht: (2025)

Your Agent Can Defend Itself against Backdoor Attacks
von: Changjiang, Li, et al.
Veröffentlicht: (2025)

Attack by Unlearning: Unlearning-Induced Adversarial Attacks on Graph Neural Networks
von: Zhang, Jiahao, et al.
Veröffentlicht: (2026)

Deciphering the Chaos: Enhancing Jailbreak Attacks via Adversarial Prompt Translation
von: Li, Qizhang, et al.
Veröffentlicht: (2024)

Systematic Scaling Analysis of Jailbreak Attacks in Large Language Models
von: Wang, Xiangwen, et al.
Veröffentlicht: (2026)

Guiding not Forcing: Enhancing the Transferability of Jailbreaking Attacks on LLMs via Removing Superfluous Constraints
von: Yang, Junxiao, et al.
Veröffentlicht: (2025)

Jailbreak Attack Initializations as Extractors of Compliance Directions
von: Levi, Amit, et al.
Veröffentlicht: (2025)

Revisiting Gradient Pruning: A Dual Realization for Defending against Gradient Attacks
von: Xue, Lulu, et al.
Veröffentlicht: (2024)

Jailbreaking Leading Safety-Aligned LLMs with Simple Adaptive Attacks
von: Andriushchenko, Maksym, et al.
Veröffentlicht: (2024)

Jailbreaking LLMs Without Gradients or Priors: Effective and Transferable Attacks
von: Nurlanov, Zhakshylyk, et al.
Veröffentlicht: (2026)

MetaCipher: A Time-Persistent and Universal Multi-Agent Framework for Cipher-Based Jailbreak Attacks for LLMs
von: Chen, Boyuan, et al.
Veröffentlicht: (2025)

CodePurify: Defend Backdoor Attacks on Neural Code Models via Entropy-based Purification
von: Mu, Fangwen, et al.
Veröffentlicht: (2024)

Jailbreaking GPT-4V via Self-Adversarial Attacks with System Prompts
von: Wu, Yuanwei, et al.
Veröffentlicht: (2023)

Revisiting Label Inference Attacks in Vertical Federated Learning: Why They Are Vulnerable and How to Defend
von: Liu, Yige, et al.
Veröffentlicht: (2026)

One Stone, Two Birds: Enhancing Adversarial Defense Through the Lens of Distributional Discrepancy
von: Zhang, Jiacheng, et al.
Veröffentlicht: (2025)