LLMSymGuard: A Symbolic Safety Guardrail Framework Leveraging Interpretable Jailbreak Concepts

@misc{hudelot2025llmsymguardsymbolicsafety,
      title={LLMSymGuard: A Symbolic Safety Guardrail Framework Leveraging Interpretable Jailbreak Concepts},
      author={Céline Hudelot and Darpan Aswal},
      year={2025},
      eprint={2508.16325},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2508.16325},
}