ResearchTasksHarmful Request DefenseHarmful Request DefenseFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedAdvBenchSafeKey0ASR65Jun 25, 2026SORRY-BenchSelf-Guard13ASR24Feb 26, 2026authority probesCognitive Firewall75Block Rate7Jul 3, 2026jbbGranite Guardian 8B0Attack Success Rate (ASR)7Jul 3, 2026