一作ACL 2026Findings· Findings of ACL
Steering Away from Refusal: A Black-box Jailbreak Method Based on First-Token Distribution
Shuangjie Fu · Du Su · Xin Chen · Fei Sun · Huawei Shen · Xueqi Cheng
摘要
Investigating black-box jailbreak attacks is crucial for revealing the actual security risks faced by operational Large Language Models (LLMs). The primary challenge in black-box jailbreak attack is the absence of direct optimization signals, such as gradients, to guide the refinement of adversar...