| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| SafeGuardPI | IBM Granite Guardian | F1 Score93 | 15 | 4d ago | |
| QualifirePI | Apriel Guard | F1 Score87 | 15 | 4d ago | |
| DeepsetPI | Apriel Guard | F1 Score71 | 15 | 4d ago | |
| Web Direct Prompt Injection | FPR0 | 7 | 4d ago | ||
| Teaching Direct Prompt Injection | FPR0 | 7 | 4d ago | ||
| Media Direct Prompt Injection | FPR0 | 7 | 4d ago | ||
| Shopping Direct Prompt Injection | FPR0 | 7 | 4d ago | ||
| Messaging Direct Prompt Injection | FPR0 | 7 | 4d ago | ||
| Language Direct Prompt Injection | AlignSentinel | FPR0 | 7 | 4d ago | |
| Entertainment Direct Prompt Injection | AlignSentinel | FPR0 | 7 | 4d ago | |
| Coding Direct Prompt Injection | FPR0 | 7 | 4d ago | ||
| AlignSentinel Evaluation Dataset (Indirect Prompt Injection Attack) | FPR (Coding)0 | 7 | 4d ago | ||
| OpenPromptInjection | Enc-first | Naive FPR0 | 6 | 4d ago | |
| IHEval Tool-use | AlignSentinel (Enc-first) | FPR0 | 6 | 4d ago | |
| IHEval Rule-following | AlignSentinel (Enc-first) | FPR0.01 | 6 | 4d ago | |
| NQ simplified | PromptArmor | Naïve RL Score6 | 5 | 4d ago | |
| AgentDojo | RedVisorLlama | Banking RL97 | 5 | 4d ago | |
| NeuralExec (test) | CONTEXTCITE | Detection Accuracy (Top-1)98.8 | 3 | 4d ago |
