List your company

Prompt-Injection Defense on Adaptive Prompt Injection (test)

37Attack Success Rate (ASR)ACT
34.4851.4968.585.51May 27, 2026

Evaluation results

MethodLinks
ACTTarget Model=GPT-OSS-20B, Attacker suffixes=best-of-82026.0537
BCTTarget Model=Phi-4-reasoning, Attacker suffixes=best-of-82026.0542.6
ACTTarget Model=Phi-4-reasoning, Attacker suffixes=best-of-82026.0549.5
BCTTarget Model=GPT-OSS-20B, Attacker suffixes=best-of-82026.0557.4
ACTTarget Model=Qwen3-8B, Attacker suffixes=best-of-82026.0559.3
ACTTarget Model=Qwen3-1.7B, Attacker suffixes=best-of-82026.0563
BCTTarget Model=Qwen3-1.7B, Attacker suffixes=best-of-82026.0580.6
BaselineTarget Model=Phi-4-reasoning, Attacker suffixes=best-of-82026.0590.9
BCTTarget Model=Qwen3-8B, Attacker suffixes=best-of-82026.0594.4
BaselineTarget Model=Qwen3-8B, Attacker suffixes=best-of-82026.0596.3
ACTTarget Model=Gemma-4-E4B-it, Attacker suffixes=best-of-82026.0596.3
BCTTarget Model=Gemma-4-E4B-it, Attacker suffixes=best-of-82026.0598.1
BaselineTarget Model=Qwen3-1.7B, Attacker suffixes=best-of-82026.05100
BaselineTarget Model=GPT-OSS-20B, Attacker suffixes=best-of-82026.05100
BaselineTarget Model=Gemma-4-E4B-it, Attacker suffixes=best-of-82026.05100