ResearchTasksDirect MaliciousFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedVLSafe OODSelfReminder90.67ASR16Apr 9, 2026MM-SafetyBench OODVLMShield0.71ASR16Apr 9, 2026