ResearchTasksMulti-agent System Safety EvaluationFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedPsySafe Dangerous Tasks 1.0w/o Defense21.5JDR R36Feb 26, 2026PsySafe Safe Tasks 1.0w/o Defense45JDR R36Feb 26, 2026