Content Safety Detection on deepset prompt-injections
93.9F1 Scoreprompt–response verification framework
Evaluation results
| Method | Links | ||
|---|---|---|---|
| prompt–response verification framework2026.06 | 93.9 | 5.3 | |
| PromptGuard-2 / InjecGuardType=Strongest Applicable Baseline2026.06 | 85.1 | — |