@inproceedings{hirota-etal-2026-guardrail,
    title = "Guardrail Models Collapse under Adversarial Paraphrases",
    author = "Hirota, Yusuke  and
      Ghosh, Shaona  and
      Wang, Yu-Chiang Frank  and
      Hachiuma, Ryo",
    editor = "Lee, Hung-Yi  and
      Dev, Sunipa  and
      Li, Jessy  and
      Elliott, Desmond",
    booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
    month = oct,
    year = "2026",
    address = "Budapest, Hungary",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2026.emnlp-main.907/",
    pages = "20895--20917",
    ISBN = "979-8-89176-459-0"
}