作者位次 #4/6ACL 2026Poster· Natural Language Processing
Between a Rock and a Hard Place: The Tension Between Ethical Reasoning and Safety Alignment in LLM s
Shei Pern Chua · Zhen Leng Thai · Kai Jun Teh · Xiao Li · Qibing Ren · Xiaolin Hu
摘要
Large Language Model safety alignment predominantly operates on a binary assumption that requests are either safe or unsafe. This classification proves insufficient when models encounter ethical dilemmas, where the capacity to reason through moral trade-offs creates a distinct attack surface. We...