Policy Violation Enforcement on MedAgentBench (Original)
0Unsafe RateHong et al. (Manual Guardrails)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Hong et al. (Manual Guardrails)Condition=Guardrail2026.06 | 0 | — | |
| Hong et al. (Manual Guardrails)Condition=Baseline2026.06 | 23 | — | |
| Hong et al. (Manual Guardrails)Condition=Raw2026.06 | 39 | — | |
| Verification Sandwich (Cedar)Condition=Raw2026.06 | — | 30.3 | |
| Verification Sandwich (Cedar)Condition=Baseline2026.06 | — | 57.7 | |
| Verification Sandwich (Cedar)Condition=Guardrail2026.06 | — | 55.7 |