ResearchTasksJailbreak Attack Safety EvaluationFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedGPT-as-a-JudgePCSA89ASR40Apr 7, 2026CARES adversarial mental-health targets 2025AMA84Success Score (SS)40Apr 7, 2026