Breaking the Shield: Adversarial Jailbreak Attacks and Defense Mechanisms in Large Language Models

Abstract

Large Language Models (LLMs) are increasingly used across diverse applications, raising security concerns, especially regarding jail-break attacks that bypass safety measures. This paper presents a systematic security evaluation of LLMs, exploring the effectiveness of jail-break detection methods, the security progression across model versions, the relationship between model size and vulnerability, and the impact of combined defense strategies. We evaluate both open-source (e.g., LLama, Mistral)

Categories

Framework mappings

Suggested from the entry's categories.

@inproceedings{dubey2025breaking,
  title = {{Breaking the Shield: Adversarial Jailbreak Attacks and Defense Mechanisms in Large Language Models}},
  author = {Shreya Dubey and H. Lamkuche},
  year = {2025},
  month = jul,
  booktitle = {2025 International Conference on Emerging Information Technology and Engineering Solutions (EITES)},
  doi = {10.1109/EITES66543.2025.00023},
  url = {https://www.semanticscholar.org/paper/01f493a008c386b93901c788bee8908b5c81a79a}
}