ResearchDatasetsCFH-HardFollowBenchmarksTask NameDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyTask NameDataset NameSOTA ResultTrendResultsAccidental DisclosureCFH-Hard Accidental89Accuracy (CFH-Hard Accidental)8Computer Use Control-Flow HijackingCFH-Hard Computer Use67Gen. Rate8Indirect Prompt Injection AttackCFH-Hard Computer Use69Attack Success Rate (IA)8