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

Adversarial Detection on Direct Harmful Prompts

100DSR

GradSafe

-4235077May 2, 2026
Updated 28d ago

Evaluation Results

MethodLinks
2026.05
100
2026.05
99.4
2026.05
99
2026.05
99
2026.05
99
2026.05
98.7
2026.05
98.3
2026.05
97.9
2026.05
96.9
2026.05
94.4
2026.05
93.4
2026.05
88.1
2026.05
81.7
2026.05
65
2026.05
58.1
2026.05
1.5
2026.05
0
2026.05
0