@inproceedings{guo-etal-2026-safety,
    title = "When Safety Routing Breaks: Understanding Alignment Fragility under Benign Fine-Tuning",
    author = "Guo, Yitong  and
      Chen, Xiaoyi  and
      Zhang, Siyuan  and
      Wang, XiaoFeng  and
      Tang, Haixu",
    editor = "Lee, Hung-Yi  and
      Dev, Sunipa  and
      Li, Jessy  and
      Elliott, Desmond",
    booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
    month = oct,
    year = "2026",
    address = "Budapest, Hungary",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2026.findings-emnlp.604/",
    pages = "12748--12762",
    ISBN = "979-8-89176-462-0"
}