Year: 2025
Type: preprint
Abstract: Jailbreak attacks have been observed to largely fail against recent reasoning models enhanced by Chain-of-Thought (CoT) reasoning. However, the underlying mechanism remains underexplored, and relying solely on reasoning capacity may raise security concerns. In this paper, we try to answer the question: Does CoT reasoning really reduce harmfulness from jailbreaking? Through rigorous theoretical analysis, we demonstrate that CoT reasoning has dual effects on jailbreaking harmfulness. Based on the theoretical insights, we propose a novel jailbreak method, FicDetail, whose practical performance validates our theoretical findings.
Source: arXiv (Cornell University)
Authors: Lu, Chengda, Fan, Xiaoyu, Huang, Yu, Xu, Rongwu, Li, Jijie, Xu, Wei
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
Awards: G366789873, G7422017672
OpenAlex ID: W6979356690
Cites:
Cited by:
Related to: