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

LivePI

Benchmarks

Task NameDataset NameSOTA ResultTrend
Indirect Prompt InjectionLivePI Total
ASR10.7
5
Indirect Prompt InjectionLivePI Gist (n=50)
ASR0
5
Indirect Prompt InjectionLivePI Repo Links (n=4)
ASR50
5
Indirect Prompt InjectionLivePI Local Docs (n=50)
ASR50
5
Indirect Prompt InjectionLivePI Email (n=50)
ASR20
5
Indirect Prompt InjectionLivePI Group chat (n=15)
ASR100
5
Showing 6 of 6 rows