LLM Stinger: Jailbreaking LLMs Using RL Fine-Tuned LLMs (Student Abstract)
Proceedings of the AAAI Conference on Artificial Intelligence · 11.04.2025 · 39 (28) · 29393-29395
DOI: 10.1609/aaai.v39i28.35263 ↗Kurz gefasst
Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 11.04.2025. Verfasst von Piyush Jha und 2 weiteren Autor:innen. Die Arbeit wurde bisher 3-mal zitiert (Crossref).
Abstract (Originalsprache)
We introduce LLM Stinger, a novel approach that leverages Large Language Models (LLMs) to automatically generate adversarial suffixes for jailbreak attacks. Unlike traditional methods, which require complex prompt engineering or white-box access, LLM Stinger uses a reinforcement learning (RL) loop to fine-tune an attacker LLM, generating new suffixes based on existing attacks for harmful questions from the HarmBench benchmark. Our method significantly outperforms existing red-teaming approaches (we compared against 15 of the latest methods), achieving a +57.2% improvement in Attack Success Rate (ASR) on LLaMA2-7B-chat and a +50.3% ASR increase on Claude 2, both models known for their extensive safety measures. Additionally, we achieved a 94.97% ASR on GPT-3.5 and 99.4% on Gemma-2B-it, demonstrating the robustness and adaptability of LLM Stinger across open and closed-source models.
Zitieren
APA
Jha, P., Arora, A., & Ganesh, V. (2025). LLM Stinger: Jailbreaking LLMs Using RL Fine-Tuned LLMs (Student Abstract). Proceedings of the AAAI Conference on Artificial Intelligence, 39(28), 29393-29395. https://doi.org/10.1609/aaai.v39i28.35263
Vancouver
Jha P, Arora A, Ganesh V. LLM Stinger: Jailbreaking LLMs Using RL Fine-Tuned LLMs (Student Abstract). Proceedings of the AAAI Conference on Artificial Intelligence. 2025;39(28):29393-29395. doi:10.1609/aaai.v39i28.35263
BibTeX
@article{Jha2025,
author = {Jha, Piyush and Arora, Arnav and Ganesh, Vijay},
title = {LLM Stinger: Jailbreaking LLMs Using RL Fine-Tuned LLMs (Student Abstract)},
journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
year = {2025},
volume = {39},
number = {28},
pages = {29393--29395},
doi = {10.1609/aaai.v39i28.35263}
}
EUVIMED