Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Safety Detection on WildJailbreak (held-out)

99AUROC

Geometry-Lite

82.8887.06591.2595.435May 18, 2026
Updated 13d ago

Evaluation Results

MethodLinks
2026.05
99
2026.05
98.4
2026.05
96.5
2026.05
92
2026.05
83.5