Jul 2026
How Jailbreak Attacks Inform Safety Alignment: A Defender-Centric, Shapley-Based Evaluation of Jailbreak Contributions
A defender-centric view of jailbreak evaluation is proposed, where attacks are evaluated by the downstream safety improvements they enable when used as red-teaming data for safety training, and A-MESS (Minimal Effective Attack-Subset Selection), a setting-agnostic framework for attributing and selecting jailbreak attacks from black-box subset utility observations is introduced.
Yukai Zhou, Feiyang Lu, Xiaokai Mao et al.
· arXiv.org · 0 citations