Defending Large Language Models Against Jailbreak Exploits with Responsible AI Considerations
Abstract
Large Language Models (LLMs) remain susceptible to jailbreak exploits that bypass safety filters and induce harmful or unethical behavior. This work presents a systematic taxonomy of existing jailbreak defenses across prompt-level, model-level, and training-time interventions, followed by three proposed defense strategies. First, a Prompt-Level Defense Framework detects and neutralizes adversarial inputs through sanitization, paraphrasing, and adaptive system guarding. Second, a Logit-Based Steering Defense reinforces refusal behavior through inference-time vector steering in safety-sensitive layers. Third, a Domain-Specific Agent Defense employs the MetaGPT framework to enforce structured, role-based collaboration and domain adherence. Experiments on benchmark datasets show substantial reductions in attack success rate, achieving full mitigation under the agent-based defense. Overall, this study highlights how jailbreaks pose a significant security threat to LLMs and identifies key intervention points for prevention, while noting that defense strategies often involve trade-offs between safety, performance, and scalability. Code is available at: https://github.com/Kuro0911/CS5446-Project
Cite
@article{arxiv.2511.18933,
title = {Defending Large Language Models Against Jailbreak Exploits with Responsible AI Considerations},
author = {Ryan Wong and Hosea David Yu Fei Ng and Dhananjai Sharma and Glenn Jun Jie Ng and Kavishvaran Srinivasan},
journal= {arXiv preprint arXiv:2511.18933},
year = {2025}
}
Comments
20 pages including appendix; technical report; NeurIPS 2024 style