List your company

Jailbreaking on EHRAgent TREQS

71.78SRNon-attack
31.68842.096552.50562.9135Apr 7, 2026

Evaluation results

MethodLinks
Non-attackAgent Backbone=Gemini-3.0-pro2026.0471.78———92.23
JailAgentAgent Backbone=Gemini-3.0-pro2026.0470.8663.9162.4761.2593.05
BadChainAgent Backbone=Gemini-3.0-pro2026.0466.1654.551.3350.9288.85
AgentPoisonAgent Backbone=Gemini-3.0-pro2026.0461.9653.4851.4349.0886.4
JailAgentAgent Backbone=GPT-4o2026.0457.8771.8869.0266.9777.2
Non-attackAgent Backbone=GPT-4o2026.0457.36———77.81
PAIRAgent Backbone=Gemini-3.0-pro2026.0455.9346.5244.2742.1380.67
AgentPoisonAgent Backbone=GPT-4o2026.0454.8172.869.2265.8573.93
BadChainAgent Backbone=GPT-4o2026.0453.1765.0361.3560.1271.68
JailAgentAgent Backbone=GPT-3.5-turbo2026.0450.164.8362.5859.260.94
Non-attackAgent Backbone=GPT-52026.0450———71.57
Non-attackAgent Backbone=GPT-3.5-turbo2026.0449.9———60.63
PAIRAgent Backbone=GPT-4o2026.0447.9653.3751.0249.5965.34
JailAgentAgent Backbone=GPT-52026.0447.3468.163.761.5571.78
AgentPoisonAgent Backbone=GPT-52026.0446.8360.9456.7553.3765.85
BadChainAgent Backbone=GPT-52026.0445.464.5262.4760.9463.7
BadChainAgent Backbone=GPT-3.5-turbo2026.0444.5861.765953.854.81
Non-attackAgent Backbone=Llama-3.1-70B2026.0444.58———82.31
JailAgentAgent Backbone=Llama-3.1-70B2026.0443.4652.0446.4245.9182.11
AgentPoisonAgent Backbone=GPT-3.5-turbo2026.0443.1560.5355.9356.1353.78
PAIRAgent Backbone=GPT-52026.0442.9455.1156.0349.961.55
AgentPoisonAgent Backbone=Llama-3.1-70B2026.0441.4151.0246.5245.479.75
PAIRAgent Backbone=GPT-3.5-turbo2026.0440.5958.1854.651.2351.33
Non-attackAgent Backbone=Claude-3.5-haiku2026.0440.18———99.49
BadChainAgent Backbone=Llama-3.1-70B2026.0439.7847.9644.2742.0276.58
JailAgentAgent Backbone=Claude-3.5-haiku2026.0439.7854.8151.7451.5399.18
AgentPoisonAgent Backbone=Claude-3.5-haiku2026.0438.3449.946.6344.5898.06
PAIRAgent Backbone=Llama-3.1-70B2026.0436.443.0539.9839.6771.88
BadChainAgent Backbone=Claude-3.5-haiku2026.0436.447.5544.1743.2596.63
PAIRAgent Backbone=Claude-3.5-haiku2026.0433.2344.8943.1542.4392.02