作者位次 #5/9ACL 2026Poster· Natural Language Processing
The “Knowledge–Behavior Gap” in Cultural Taboo Safety of Large Language Models
Ying He · Sihang Jiang · Xingzhou Chen · Zhouhong Gu · Yiwei Gu · Minggui HE · Shimin Tao · Mahongxia · Yanghua Xiao
摘要
Cultural taboo safety is essential for deploying large language models (LLMs), as culturally insensitive outputs may cause offense or even social harm. However, existing cultural benchmarks primarily assess cultural knowledge or values biases, while overlooking whether LLMs can recognize and resp...