| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| AdvBench | AMIS | ASR100 | 132 | 3mo ago | |
| HarmBench | MASKFORGE | Attack Success Rate (ASR)84 | 110 | 1mo ago | |
| AdvBench selected models | Few-shot threat model | ASR@10100 | 90 | 2mo ago | |
| AdvBench | GPTFuzzer | ASR0 | 88 | 29d ago | |
| AdvBench | JULI | BERT Score4.84 | 55 | 4mo ago | |
| HARMBENCH 159 standard behaviors (test) | ASR0 | 55 | 1mo ago | ||
| JailbreakBench | Attack Success Rate (ASR)2 | 53 | 2mo ago | ||
| MHSC | TCBS-Attack | ASR-431 | 44 | 4mo ago | |
| Q16 | TCBS-Attack | ASR-452.5 | 44 | 4mo ago | |
| JailbreakBench | MASKFORGE | ASR85 | 42 | 1mo ago | |
| StrongReject (test) | PAD | Attack Success Rate (ASR)89.8 | 42 | 1mo ago | |
| AdvBench Sub | BERT Score4.73 | 40 | 4mo ago | ||
| JBB Behaviors | AMIS | ASR100 | 35 | 3mo ago | |
| AdvBench (test) | CHaRS-PCT | Average ASR99.04 | 33 | 4mo ago | |
| MaliciousInstruct (test) | JULI | BERT Score4.63 | 32 | 4mo ago | |
| EHRAgent TREQS | SR71.78 | 30 | 3mo ago | ||
| EHRAgent eICU | Success Rate (SR)57.93 | 30 | 3mo ago | ||
| EHRAgent MIMIC-III | SR56.55 | 30 | 3mo ago | ||
| JBB-Behaviors (test) | HMNS | ASR (GPT-4o)99 | 27 | 3mo ago | |
| HarmBench (test) | HMNS | ASR (GPT-4o)97 | 27 | 22d ago | |
| AdvBench (test) | HMNS | ASR (GPT-4o)99 | 27 | 3mo ago | |
| AdvBench 20% evaluation | A-LQR+ | ASR97.12 | 25 | 3mo ago | |
| EHRAgent ALL | JailAgent | Weighted Average ASR70.112 | 24 | 3mo ago | |
| ConceptRisk | VII | Attack Success Rate68 | 24 | 4mo ago | |
| COCO-I2VSafetyBench | ASR17 | 24 | 4mo ago |