机构
*
James Cook University(詹姆斯库克大学)
;
Western Sydney University(西悉尼大学)
;
Adelaide University(阿德莱德大学)
;
CSIRO(联邦科学与工业研究组织)
;
The University of Osaka(大阪大学)
;
Nanjing University of Science and Technology(南京理工大学)
;
Macquarie University(麦考瑞大学)
;
La Trobe University(拉筹伯大学)
机构
*
University of Notre Dame(notre dame 大学)
;
University of Washington(华盛顿大学)
;
Johns Hopkins University(约翰霍普金斯大学)
;
Georgia Institute of Technology(佐治亚理工学院)
Comments17 pages, 7 figures, 9 tables. Code and data available at this https URL (https://github.com/erikl2/safety-refusal-stability). v3: corrects dataset attribution (AdvBench+HarmBench, not BeaverTails), fixes label-derived statistics and CIs, revises the judge-noise analysis to a sensitivity framing; conclusions unchanged
LongGuard: Mechanistic Analysis and Training-Free Mitigation of Long-Context Failure in Safety Guardrails
LongGuard:针对安全护栏长上下文失效的机制分析与无训练缓解方案
Ziyang Chen, Xing Wu, Songlin Hu
机构
*
Institute of Information Engineering, Chinese Academy of Sciences(中国科学院信息工程研究所)
;
School of Cyber Security, University of Chinese Academy of Sciences(中国科学院大学网络空间安全学院)
机构
*
School of Cyber Science and Engineering, Huazhong University of Science and Technology(华中科技大学网络空间科学与工程学院)
;
School of Mathematical Sciences, Peking University(北京大学数学科学学院)
;
School of Software and engineering, Huazhong University of Science and Technology(华中科技大学软件学院)
;
School of Computer Science, Nanjing University(南京大学计算机学院)
CommentsAccepted at the 2nd Workshop on Reliable and Responsible Foundation Models of the 42nd International Conference on Machine Learning (ICML) 2025
机构
*
University of Chinese Academy of Sciences(中国科学院大学)
;
Nanyang Technological University(南洋理工大学)
;
Supply Chain Tech Team Y, JD.com(京东Y供应链技术团队)
;
Peking University(北京大学)
;
Fudan University(复旦大学)
;
Fullive-AI(Fullive人工智能)