TADSOB: Enhancing jailbreaking attacks on LLMs via trajectory-aware dynamic suffix optimization with buffer

Neurocomputing 2026 Open access

Read full text

Authors: Zhiyi Liu, Mingrui Lao, Chengsi Du, Yunhao Feng, Xiaohu Du, Liang Bai, Yu Guo

DOI: 10.1016/j.neucom.2026.135010

Cited by: 2

References: 1

Fields of Study & Subjects

Adversarial Robustness in Machine Learning Topic Modeling Natural Language Processing Techniques

References 1