ResearchTasksSafety RefusalSafety RefusalFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedAdvBenchLow-Rank Combination99.42Refusal Rate46Mar 17, 2026RefusalLCF0ASR42Apr 28, 2026Safety Evaluation PromptsDirAbl62.5Refusal40Feb 26, 2026Jailbreak PromptsITI87.2Refusal Rate15Mar 17, 2026ToxicChatWAS95Refusal Rate15Mar 17, 2026