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

Malicious Prompt Detection on JailbreakV_28K Image-based (test)

0.19FNR

VLMShield

-3.00418.555540.11561.6745Apr 7, 2026
Updated 9d ago

Evaluation Results

MethodLinks
2026.04
0.19
2026.04
0.19
2026.04
2.14
2026.04
5.21
2026.04
11.82
2026.04
14
2026.04
16.37
2026.04
17.19
2026.04
17.19
2026.04
22.05
2026.04
34.8
2026.04
37.2
2026.04
37.2
2026.04
39.68
2026.04
43.06
2026.04
80.04