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

Indirect Prompt Injection on LLM Behavior Subset 1

99.8IR

Actor-Critic

-3.99222.95449.976.846May 14, 2026
Updated 16d ago

Evaluation Results

MethodLinks
2026.05
99.8
2026.05
99.4
2026.05
96.2
2026.05
95
2026.05
88.6
2026.05
86.4
2026.05
80.4
2026.05
79
2026.05
78.8
2026.05
75
2026.05
64.2
2026.05
62.8
2026.05
62.4
2026.05
54.2
2026.05
13.6
2026.05
5.4
2026.05
4.2
2026.05
3.8
2026.05
3
2026.05
2.6
2026.05
2.2
2026.05
2
2026.05
0.2
2026.05
0