Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Indirect Prompt Injection Defense on AgentDojo Banking and Slack suites
Loading...
35.29
BU (No Attack)
No Defense
16.9444
21.7072
26.47
31.2328
Mar 11, 2026
BU (No Attack)
Utility Accuracy (Important Messages)
Attack Success Rate (Important Messages)
Utility Accuracy (Tool Knowledge)
Attack Success Rate (Tool Knowledge)
Updated 1mo ago
Evaluation Results
Method
Method
Links
BU (No Attack)
Utility Accuracy (Important Messages)
Attack Success Rate (Important Messages)
Utility Accuracy (Tool Knowledge)
Attack Success Rate (Tool Knowledge)
No Defense
Model=Llama3-8B
2026.03
35.29
35.07
10.9
35.55
17.06
AttriGuard
Model=Llama3-8B
2026.03
32.35
31.75
0
26.07
0
SecAlign
Model=Llama3-8B
2026.03
23.53
22.75
0
23.22
0
StruQ
Model=Llama3-8B
2026.03
17.65
23.22
0
23.22
0
Feedback
Search any
task
Search any
task