List your company

Prompt Injection Attack on AgentDojo 13 non-agent benchmarks

0Training QueriesTAP
-1,2006,90015,00023,100Mar 13, 2026

Evaluation results

MethodLinks
TAPTarget Model=Meta-SecAlign-8B, Evaluation Setting=ASR@102026.03082,11082,110
PAIRTarget Model=Meta-SecAlign-8B, Evaluation Setting=ASR@102026.03074,46074,460
StrategyTarget Model=Meta-SecAlign-8B, Evaluation Setting=ASR@102026.030194,140194,140
Vanilla GRPOTarget Model=Meta-SecAlign-8B, Evaluation Setting=ASR@102026.0315,00017,00032,000
PISmithTarget Model=Meta-SecAlign-8B, Evaluation Setting=ASR@102026.0315,00017,00032,000
RL-HammerTarget Model=Meta-SecAlign-8B, Evaluation Setting=ASR@102026.0330,00017,00047,000