一作ACL 2026Findings· Findings of ACL
Rethinking Reasoning: A Survey on Reasoning-based Backdoors in LLM s
Man Hu · Xinyi Wu · Zhufeng Suo · Jinbo Feng · Linghui Meng · Yanhao Jia · Anh Tuan Luu · Shuai Zhao
摘要
With the rise of advanced reasoning capabilities, large language models (LLMs) are receiving increasing attention. While reasoning enhances LLMs’ performance on downstream tasks, it also introduces new threat vectors, as adversaries can leverage these capabilities to conduct backdoor attacks. Pri...