ResearchTasksJailbreak attacksFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedMM-SafetyTRR99.9Safety Rate22Jul 7, 2026MML-Mirror (MML-M)TRR97Safety Rate12Feb 26, 2026