ResearchTasksMultimodal Large Language Model Safety EvaluationMultimodal Large Language Model Safety EvaluationFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedMM-SafetyBench++GPT-5100Illegal Activity Unsafe Refusal Rate9Mar 18, 2026