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

Jailbreaking Attack on MM-SafetyBench

95Attack Success Rate (ASR)

TAO-Attack

-3.404822.142647.6973.2374Jan 8, 2026Jan 29, 2026Feb 20, 2026Mar 14, 2026Apr 4, 2026Apr 26, 2026May 18, 2026
Updated 14d ago

Evaluation Results

MethodLinks
2026.03
95132
2026.05
93.75-
2026.05
93.1-
2026.01
91.5-
2026.05
91.25-
2026.05
90.42-
2026.03
90248
2026.05
88.45-
2026.05
86.24-
2026.05
82.86-
2026.05
82.02-
2026.05
80.65-
2026.05
80.54-
2026.05
79.4-
2026.01
77.31-
2026.01
72.14-
2026.01
68.57-
2026.01
62.68-
2026.01
62.3-
2026.05
29.4-
2026.05
28.93-
2026.05
28.27-
2026.05
26.9-
2026.05
25.48-
2026.05
21.3-
2026.01
19.81-
2026.05
19.29-
2026.05
16.49-
2026.05
15.24-
2026.01
12.34-
2026.05
11.73-
2026.05
11.31-
10-
2026.04
7.31-
2026.04
2.69-
2026.04
2.69-
1.54-
0.77-
0.77-
2026.04
0.77-
0.77-
2026.04
0.38-