New Benchmark Reveals Frontier AI Models Struggle with Multi-Level Instruction Conflicts
Research shows current leading AI models achieve only ~40% accuracy when handling conflicts among 12 levels of instructions in agent settings.
Sources · arxiv.org