יום שישי, 31 ביולי 2026 LIVE
AI־INFO

כתבה arXiv cs.CL ·

Language Shapes Instruction Hierarchy Compliance in Multilingual LLMs

תקציר מקורי באנגליתarXiv:2607.23545v1 Announce Type: new Abstract: Instruction hierarchy (IH) requires models to prioritize instructions by source, ensuring that higher-priority instructions override lower-priority ones. Despite its importance for safe and controllable deployment, existing evaluations have focused almost exclusively on English, leaving it unclear whether IH compliance remains stable in multilingual settings. We introduce XIH-Bench, a benchmark for multilingual IH evaluation with both same-language and cross-language conflicts across six languages, four domains, and three IH settings. Across models, we find two consistent patterns. First, IH compliance exhibits a clear language-dependent asymmetry: a language that strengthens compliance in the higher-priority position can become disruptive in
קרא במקור המקורי