Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Prompt-Injection Defense on Adaptive Prompt Injection (test)
Loading...
37
Attack Success Rate (ASR)
ACT
34.48
51.49
68.5
85.51
May 27, 2026
Attack Success Rate (ASR)
Updated 6d ago
Evaluation Results
Method
Method
Links
Attack Success Rate (ASR)
ACT
Target Model=GPT-OSS-2...
2026.05
37
BCT
Target Model=Phi-4-rea...
2026.05
42.6
ACT
Target Model=Phi-4-rea...
2026.05
49.5
BCT
Target Model=GPT-OSS-2...
2026.05
57.4
ACT
Target Model=Qwen3-8B,...
2026.05
59.3
ACT
Target Model=Qwen3-1.7...
2026.05
63
BCT
Target Model=Qwen3-1.7...
2026.05
80.6
Baseline
Target Model=Phi-4-rea...
2026.05
90.9
BCT
Target Model=Qwen3-8B,...
2026.05
94.4
Baseline
Target Model=Qwen3-8B,...
2026.05
96.3
ACT
Target Model=Gemma-4-E...
2026.05
96.3
BCT
Target Model=Gemma-4-E...
2026.05
98.1
Baseline
Target Model=Qwen3-1.7...
2026.05
100
Baseline
Target Model=GPT-OSS-2...
2026.05
100
Baseline
Target Model=Gemma-4-E...
2026.05
100
Feedback
Search any
task
Search any
task