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

LLM Behavior

Benchmarks

Task NameDataset NameSOTA ResultTrend
Indirect Prompt InjectionLLM Behavior Subset 1
IR99.8
24
LLM BehaviorLLM Behavior
Response Rate (RR)95
12
Prompt Injection Attack SuccessLLM Behavior
Injection Rate (IR)100
10
Indirect Prompt Injection Attack Success EvaluationLLM Behavior Goal-Distant
IRany100
5
Indirect Prompt Injection Attack Success EvaluationLLM Behavior Goal-Adjacent
IRany100
5
Showing 5 of 5 rows