ResearchDatasetsUltraChat StrongREJECTFollowBenchmarksTask NameDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyTask NameDataset NameSOTA ResultTrendResultsDirectional Jailbreak DetectionUltraChat (Safe) × StrongREJECT (Harmful)0.797AUROC (mono)3