@inproceedings{chu-etal-2026-detection,
    title = "From Detection to Refusal: Safer {LLM}s via Circuit-Guided Weight Scaling",
    author = "Chu, Kuan-Lin  and
      Sun, Chung-En  and
      Weng, Tsui-Wei",
    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.1629/",
    doi = "10.18653/2026.findings-emnlp.1629",
    pages = "34657--34680",
    ISBN = "979-8-89176-462-0"
}