[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-seo-273406-105":3,"detail-sidebar-cat-0-en-105":81,"doc-detail-273406-en":130},{"code":4,"msg":5,"data":6},0,"ok",{"site_id":7,"language":8,"slug":9,"title":10,"keywords":11,"description":12,"schema_data":13,"social_meta":74,"head_meta":76,"extra_data":78,"updated_unix":80},105,"en","language-shapes-instruction-hierarchy-compliance-in-multilingual-llms","Language Shapes Instruction Hierarchy Compliance in Multilingual LLMs","","Instruction hierarchy (IH) requires models to prioritize instructions by source so higher-priority inputs override lower-priority ones. Despite its role in safe, controllable deployment, most evaluations focus on English, leaving uncertainty about multilingual stability. This work introduces XIH-Bench, enabling multilingual IH assessment with same-language and cross-language conflicts across six languages, four domains, and three IH settings. Experiments reveal language-dependent asymmetry and the Language Boundary Effect.",{"@graph":14,"@context":73},[15,34,56],{"@type":16,"itemListElement":17},"BreadcrumbList",[18,23,27,31],{"item":19,"name":20,"@type":21,"position":22},"https://docshare.wps.com","Home","ListItem",1,{"item":24,"name":25,"@type":21,"position":26},"https://docshare.wps.com/document/","Document",2,{"item":28,"name":29,"@type":21,"position":30},"https://docshare.wps.com/document/research-report/","Research & Report",3,{"item":32,"name":10,"@type":21,"position":33},"https://docshare.wps.com/document/language-shapes-instruction-hierarchy-compliance-in-multilingual-llms/273406/",4,{"url":32,"name":10,"@type":35,"image":36,"author":41,"headline":10,"publisher":44,"fileFormat":47,"inLanguage":8,"description":12,"dateModified":48,"datePublished":49,"encodingFormat":47,"isAccessibleForFree":50,"interactionStatistic":51},"DigitalDocument",{"url":37,"@type":38,"width":39,"height":40},"https://docshare.wps.com/thumbnails/language-shapes-instruction-hierarchy-compliance-in-multilingual-llms/273406.png","ImageObject",300,407,{"name":42,"@type":43},"Anda","Person",{"url":19,"name":45,"@type":46},"DocShare","Organization","application/pdf","2026-11-01","2026-09-15",true,{"@type":52,"interactionType":53,"userInteractionCount":55},"InteractionCounter",{"@type":54},"ViewAction",10,{"@type":57,"mainEntity":58},"FAQPage",[59,65,69],{"name":60,"@type":61,"acceptedAnswer":62},"What problem does instruction hierarchy address in LLM deployments?","Question",{"text":63,"@type":64},"Instruction hierarchy determines which instructions take precedence when multiple sources coexist in the same context. Higher-priority sources should override lower-priority ones to ensure correct behavior under conflicts.","Answer",{"name":66,"@type":61,"acceptedAnswer":67},"What is XIH-Bench and what does it evaluate?",{"text":68,"@type":64},"XIH-Bench is a multilingual benchmark for instruction-hierarchy evaluation. It tests both same-language and cross-language conflicts while varying the languages assigned to higher- and lower-priority instructions across multiple domains and settings.",{"name":70,"@type":61,"acceptedAnswer":71},"What are the main findings about multilingual IH compliance?",{"text":72,"@type":64},"The study finds clear language-dependent asymmetry, where the same language can help in one IH position but harm in the other. It also observes that cross-language conflicts yield higher compliance than same-language conflicts, termed the Language Boundary Effect.","https://schema.org",{"og:url":32,"og:type":75,"og:title":10,"og:site_name":45,"og:description":12},"article",{"robots":77,"canonical":32},"index,follow",{"doc_id":79,"site_id":7},273406,1789471264,{"code":4,"msg":82,"data":83},"success",[84,88,92,96,101,106,111,115,120,123,126],{"id":22,"doc_module":4,"doc_module_name":25,"category_name":85,"show_sort_weight":86,"slug":87},"Story & Novel",90,"story-novel",{"id":26,"doc_module":4,"doc_module_name":25,"category_name":89,"show_sort_weight":90,"slug":91},"Literature",80,"literature",{"id":33,"doc_module":4,"doc_module_name":25,"category_name":93,"show_sort_weight":94,"slug":95},"Exam",70,"exam",{"id":97,"doc_module":4,"doc_module_name":25,"category_name":98,"show_sort_weight":99,"slug":100},5,"Comic",60,"comic",{"id":102,"doc_module":4,"doc_module_name":25,"category_name":103,"show_sort_weight":104,"slug":105},6,"Technology",50,"technology",{"id":107,"doc_module":4,"doc_module_name":25,"category_name":108,"show_sort_weight":109,"slug":110},7,"Healthcare",40,"healthcare",{"id":112,"doc_module":4,"doc_module_name":25,"category_name":29,"show_sort_weight":113,"slug":114},8,30,"research-report",{"id":116,"doc_module":4,"doc_module_name":25,"category_name":117,"show_sort_weight":118,"slug":119},9,"Religion & Spirituality",20,"religion-spirituality",{"id":118,"doc_module":4,"doc_module_name":25,"category_name":121,"show_sort_weight":118,"slug":122},"World Cup","world-cup",{"id":55,"doc_module":4,"doc_module_name":25,"category_name":124,"show_sort_weight":55,"slug":125},"Lifestyle","lifestyle",{"id":127,"doc_module":4,"doc_module_name":25,"category_name":128,"show_sort_weight":97,"slug":129},19,"General","general",{"code":4,"msg":82,"data":131},{"doc_id":79,"user_id":132,"nickname":42,"user_avatar":133,"doc_module":4,"category_id":112,"category_name":29,"doc_title":10,"doc_description":12,"doc_content":134,"file_id":135,"file_url":136,"file_type":137,"file_size":138,"view_count":55,"is_deleted":4,"is_public":22,"is_downloadable":22,"audit_status":22,"page_count":139,"language":140,"language_code":8,"site_id":7,"html_lang":8,"table_of_contents":141,"faqs":142,"seo_title":143,"seo_description":12,"update_tm":80,"read_time":144,"preview_page_count":4},962075006959,"https://ap-avatar.wpscdn.com/avatar/e0002397efbe92a78e?_k=1776741047341049297","Language Shapes Instruction Hierarchy Compliance in Multilingual LLMs  \nJiwon Moon1 Yerin Hwang1 Kyomin Jung1,2†  \n1IPAI, Seoul National University 2Dept. of ECE, Seoul National University  \n{wldnjs913, dpfls589, [kjung}@snu.ac.kr](kjung}@snu.ac.kr)  \narXiv : 2607 .23545v2 [cs .CL] 2 1 Aug 2026  \nAbstract  \nInstruction hierarchy (IH) requires models to prioritize instructions by source, ensuring that higher-priority instructions override lowerpriority 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.  \nAcross models, we find two consistent patterns.  \nFirst, IH compliance exhibits a clear languagedependent asymmetry: a language that strengthens compliance in the higher-priority position can become disruptive in the lower-priority position. Second, cross-language conflicts yield higher compliance than same-language conflicts, a phenomenon we term the Language Boundary Effect. We further show that language specialization can make lower-priority instructions in model-favored languages harder to override, creating multilingual reliability and security risks. Code and data are available at [github.com/g1moon/Language-Shapes-IH](github.com/g1moon/Language-Shapes-IH).  \n1 Introduction  \nLarge language models (LLMs) are increasingly deployed in contexts where instructions from multiple sources coexist within a single context window, including system prompts, user messages, and tool outputs (Li et al., 2024a ; Debenedetti et al., 2024 ; Zhang et al., 2025a) . Correct behavior in these multi-source settings depends not only on whether a model follows an instruction, but on whether it follows the right instruction when sources conflict. This requirement is commonly formalized as an instruction hierarchy (IH) (Wallace et al., 2024 ; Guo et al., 2026), under which higher-priority sources should override lower-priority ones. Preserving  \n† Corresponding author  \nIncluez le titre entre \u003C\u003C >>.  \n(=Include the title in \u003C\u003C >>.)  \nUSER MESSAGE  \nLOW PRIORITY  \nएक $ब&ली के बारे म- एक क$वता 0ल1खए।  \nशीषHक को { } के भीतर शाMमल कर- ।(=Write a poem about a cat.  \nInclude the title in { }.)  \nMODEL RESPONSE  \n\u003C\u003C The Midnight Explorer >>  \nSoft paws tread on the hardwood floor …  \nExpected: System > User  \nHIERARCHYCOMPLIANT  \nMODEL RESPONSE  \n{ The Midnight Explorer }  \nSoft paws tread on the hardwood floor …  \nObserved: System \u003C User  \nHIERARCHYVIOLATING  \nFigure 1: Multilingual instruction hierarchy conflict under the Sys>User setting. The same conflict is resolved correctly in one language configuration (left) but incorrectly in another (right), showing language-dependent variation in hierarchy compliance.  \nIH is essential for safe and controllable deployment, since failures can enable jailbreaks, prompt injection, and system prompt leakage (Zou et al., 2023 ; Liu et al., 2024 ; Agarwal et al., 2024) . This challenge is especially salient in multilingual deployments, where higher- and lower-priority instructions may be written in different languages, allowing a lower-priority instruction in one language to interfere with or even override a higher-priority instruction in another (Figure 1) .  \nUnder IH, priority is determined by source rather than language, so IH compliance should in principle be invariant across languages. However, existing IH benchmarks are almost entirely Englishcentric (Wu et al., 2024 ; Zhang et al., 2025b), leaving it unclear whether source conflicts are resolved consistently across languages. Although multilingual instruction-following research has shown substantial cross-lingual variation (He et al., 2024 ; Zeng et al., 2025 ; Li et al., 2025), it mainly examines compliance with a single","cbCairzZBSwP5FmH","https://ap.wps.com/l/cbCairzZBSwP5FmH","pdf",1009195,33,"English","# Introduction\n## Instruction hierarchy and multilingual settings\n## XIH-Bench: benchmark design\n## Key findings and implications","[{\"question\":\"What problem does instruction hierarchy address in LLM deployments?\",\"answer\":\"Instruction hierarchy determines which instructions take precedence when multiple sources coexist in the same context. Higher-priority sources should override lower-priority ones to ensure correct behavior under conflicts.\"},{\"question\":\"What is XIH-Bench and what does it evaluate?\",\"answer\":\"XIH-Bench is a multilingual benchmark for instruction-hierarchy evaluation. It tests both same-language and cross-language conflicts while varying the languages assigned to higher- and lower-priority instructions across multiple domains and settings.\"},{\"question\":\"What are the main findings about multilingual IH compliance?\",\"answer\":\"The study finds clear language-dependent asymmetry, where the same language can help in one IH position but harm in the other. It also observes that cross-language conflicts yield higher compliance than same-language conflicts, termed the Language Boundary Effect.\"}]","Language Shapes Instruction Hierarchy Compliance in Multilingual LLMs | PDF",83]