ResearchTasksSafety Alignment Robustness EvaluationFor AI assistantsMarkdownContent indexSafety Alignment Robustness EvaluationFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedPKU-SafeRLHF (n=100 samples)SFT Base18.5Random Rate10May 12, 2026