Voices Across Registers: Corpus-Conditioned Vernacular Jailbreaks against Aligned LLMs via Fanfiction Subgenres
分布外声音:同人小说子类型作为对齐LLM的通用白话越狱
机构 * School of Science and Engineering, The Chinese University of Hong Kong (Shenzhen)(香港中文大学(深圳)科学与工程学院) ; The Shenzhen Future Network of Intelligence Institute (FNii-Shenzhen)(深圳未来网络智能研究院) ; The Guangdong Provincial Key Laboratory of Future Networks of Intelligence(广东省未来网络智能重点实验室) ; School of Microelectronics, Xi’an Jiaotong University(西安交通大学微电子学院)
AI总结 本文发现安全训练覆盖不足的自然人类写作语域是对齐LLM的真正失败模式,并提出首个利用真实同人小说子类型作为通用攻击载体的越狱方法,显著提升攻击成功率。
Comments Accepted by EMNLP 2026 Main Conference, 25 pages