Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Indirect Prompt Injection Defense on AgentDojo (Full)
Loading...
78.26
BU (NOATTACK)
MetaSecAlign
53.2896
59.7723
66.255
72.7377
Mar 11, 2026
BU (NOATTACK)
UA (IMPORTANTMSGS)
ASR (IMPORTANTMSGS)
UA (TOOLKNOWLEDGE)
ASR (TOOLKNOWLEDGE)
Updated 1mo ago
Evaluation Results
Method
Method
Links
BU (NOATTACK)
UA (IMPORTANTMSGS)
ASR (IMPORTANTMSGS)
UA (TOOLKNOWLEDGE)
ASR (TOOLKNOWLEDGE)
MetaSecAlign
Model=Llama3.3-70B
2026.03
78.26
78.91
0.79
77.32
0.79
No Defense
Model=Llama3.3-70B
2026.03
59.78
43.08
22.11
47.51
18.82
AttriGuard
Model=Llama3.3-70B
2026.03
54.25
47.62
0
45.8
0
Feedback
Search any
task
Search any
task