[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-273688-105":59,"doc-detail-273688-en":130},{"code":4,"msg":5,"data":6},0,"success",[7,13,18,23,28,33,38,43,48,51,55],{"id":8,"doc_module":4,"doc_module_name":9,"category_name":10,"show_sort_weight":11,"slug":12},1,"Document","Story & Novel",90,"story-novel",{"id":14,"doc_module":4,"doc_module_name":9,"category_name":15,"show_sort_weight":16,"slug":17},2,"Literature",80,"literature",{"id":19,"doc_module":4,"doc_module_name":9,"category_name":20,"show_sort_weight":21,"slug":22},4,"Exam",70,"exam",{"id":24,"doc_module":4,"doc_module_name":9,"category_name":25,"show_sort_weight":26,"slug":27},5,"Comic",60,"comic",{"id":29,"doc_module":4,"doc_module_name":9,"category_name":30,"show_sort_weight":31,"slug":32},6,"Technology",50,"technology",{"id":34,"doc_module":4,"doc_module_name":9,"category_name":35,"show_sort_weight":36,"slug":37},7,"Healthcare",40,"healthcare",{"id":39,"doc_module":4,"doc_module_name":9,"category_name":40,"show_sort_weight":41,"slug":42},8,"Research & Report",30,"research-report",{"id":44,"doc_module":4,"doc_module_name":9,"category_name":45,"show_sort_weight":46,"slug":47},9,"Religion & Spirituality",20,"religion-spirituality",{"id":46,"doc_module":4,"doc_module_name":9,"category_name":49,"show_sort_weight":46,"slug":50},"World Cup","world-cup",{"id":52,"doc_module":4,"doc_module_name":9,"category_name":53,"show_sort_weight":52,"slug":54},10,"Lifestyle","lifestyle",{"id":56,"doc_module":4,"doc_module_name":9,"category_name":57,"show_sort_weight":24,"slug":58},19,"General","general",{"code":4,"msg":60,"data":61},"ok",{"site_id":62,"language":63,"slug":64,"title":65,"keywords":66,"description":67,"schema_data":68,"social_meta":123,"head_meta":125,"extra_data":127,"updated_unix":129},105,"en","auditing-cross-lingual-fairness-in-language-model-watermarking","Auditing Cross-Lingual Fairness in Language Model Watermarking","","Watermarking schemes for large language model output are evaluated largely on English using each scheme’s detection threshold and a limited set of quality metrics. Multilingual deployment makes evaluation-design choices decisive for conclusions. The work introduces a four-part evaluation framework: empirically calibrated thresholds by deployment context, a threshold-independent companion metric separating calibration from detection failures, three disjoint quality paradigms, and a generalized-entropy decomposition across typological partitions. Applied to multiple schemes, generators, languages, and regimes, it exposes fairness failure modes not detectable by single-language evaluations.",{"@graph":69,"@context":122},[70,84,105],{"@type":71,"itemListElement":72},"BreadcrumbList",[73,77,79,82],{"item":74,"name":75,"@type":76,"position":8},"https://docshare.wps.com","Home","ListItem",{"item":78,"name":9,"@type":76,"position":14},"https://docshare.wps.com/document/",{"item":80,"name":40,"@type":76,"position":81},"https://docshare.wps.com/document/research-report/",3,{"item":83,"name":65,"@type":76,"position":19},"https://docshare.wps.com/document/auditing-cross-lingual-fairness-in-language-model-watermarking/273688/",{"url":83,"name":65,"@type":85,"image":86,"author":91,"headline":65,"publisher":94,"fileFormat":97,"inLanguage":63,"description":67,"dateModified":98,"datePublished":99,"encodingFormat":97,"isAccessibleForFree":100,"interactionStatistic":101},"DigitalDocument",{"url":87,"@type":88,"width":89,"height":90},"https://docshare.wps.com/thumbnails/auditing-cross-lingual-fairness-in-language-model-watermarking/273688.png","ImageObject",300,407,{"name":92,"@type":93},"River Wang","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-10-29","2026-09-15",true,{"@type":102,"interactionType":103,"userInteractionCount":52},"InteractionCounter",{"@type":104},"ViewAction",{"@type":106,"mainEntity":107},"FAQPage",[108,114,118],{"name":109,"@type":110,"acceptedAnswer":111},"Why is cross-lingual fairness a concern in language model watermarking evaluations?","Question",{"text":112,"@type":113},"Most watermark evaluations rely on English prompts and English text with scheme-specific detection thresholds. In multilingual settings, evaluation-design choices that seem irrelevant in English can substantially change conclusions across languages.","Answer",{"name":115,"@type":110,"acceptedAnswer":116},"What is the proposed evaluation framework for multilingual watermarking?",{"text":117,"@type":113},"It combines empirically calibrated detection thresholds, a threshold-independent companion metric to separate calibration failures from detection failures, three distinct quality measurement paradigms, and a generalized-entropy decomposition of cross-language disparity over a typological partition.",{"name":119,"@type":110,"acceptedAnswer":120},"What do the experiments reveal about where fairness disparities originate?",{"text":121,"@type":113},"Disparity is typically large, scheme-specific, and predominantly between-family on the typological partition. This suggests watermarking cross-lingual fairness gaps are structural to language properties rather than quirks of particular languages.","https://schema.org",{"og:url":83,"og:type":124,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":126,"canonical":83},"index,follow",{"doc_id":128,"site_id":62},273688,1789472500,{"code":4,"msg":5,"data":131},{"doc_id":128,"user_id":132,"nickname":92,"user_avatar":133,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":67,"doc_content":134,"file_id":135,"file_url":136,"file_type":137,"file_size":138,"view_count":52,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":139,"language":140,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":141,"faqs":142,"seo_title":143,"seo_description":67,"update_tm":129,"read_time":26,"preview_page_count":4},1099514067438,"https://ap-avatar.wpscdn.com/avatar/100002539ee87300030?x-image-process=image/resize,m_fixed,w_180,h_180&k=1780474512215547542","arXiv : 2608 .20047v 1 [cs .CL] 20 Aug 2026  \nAuditing Cross-Lingual Fairness in Language Model  \nWatermarking  \nAlexander NemecekOsama Zafar, Debargha Ganguly, Vikash Singh,  \nVipin Chaudhary, Erman Ayday  \nCase Western Reserve University  \n{ajn98,oxz23,dxg512,vxs465,vxc204,[exa208}@case.edu](exa208}@case.edu)  \nAbstract  \nWatermarking schemes for large language model output are evaluated almost exclusively on English text using each scheme’s detection threshold and a narrow set of quality measurements. Multilingual deployment exposes evaluation-design choices that are inconsequential on English but determine conclusions cross-lingually. We propose an evaluation framework with four components: detection thresholds calibrated empirically per deployment context, a threshold-independent companion measurement that distinguishes calibration failures from detection failures, three disjoint quality measurement paradigms (distributional, paired-semantic, and reference-perplexity), and a generalized-entropy decomposition of cross-language disparity over a typological family partition. Applied to six watermarking schemes, three open-weight generators, eleven languages spanning four scripts and eight typological families, and both base and instruction-tuned regimes, the framework reveals failure modes that single-language single-paradigm evaluation cannot surface.  \nAcross detection and quality, observed disparity is predominantly between-family on the typological partition, indicating that cross-lingual fairness gaps in watermarking are structural to language properties rather than idiosyncratic to particular languages.  \n1 Introduction  \nRecent advances in large language models (LLMs) have made AI-generated text difficult to distinguish from human-authored writing. This raises concerns about misinformation, the erosion of trust in online content, and model collapse, in which the performance of models trained on partially machinegenerated data degrades over successive generations [1, 24, 22] . The misinformation concern in particular is not hypothetical. NewsGuard has identified over 3,000 AI-generated news and information sites publishing in 16 languages, often under legitimate-sounding names and with little to no human oversight [20] .  \nText watermarking offers one mitigation by embedding a recoverable statistical signature into model output. A watermarking scheme modifies the generation process in a key-dependent way, most commonly by biasing the next-token distribution toward a pseudorandomly selected vocabulary subset [10, 15, 2] . A paired detector then tests a candidate text against this signature and returns ascore, which is thresholded to yield a detection decision [6] . Schemes are typically evaluated on detectability, generation quality (fluency, coherence, and distributional similarity to unwatermarked output), and robustness to downstream perturbations [18, 25, 12] .  \nThese evaluations, however, are conducted almost exclusively on English prompts and English generated text. The implicit assumption is that a scheme which detects reliably and preserves quality  \n∗Corresponding author.  \nPreprint.  \non English will do the same on other languages, but this assumption has not been empirically tested at scale. This gap matters in practice, because watermarking is already deployed in settings where users prompt and generate in dozens of languages [2, 14], and the multilingual misinformation ecosystem flagged above is precisely the threat surface that motivates the technology in the first place.  \nOur work. We present the first systematic audit of cross-lingual fairness in LLM text watermarking. Our contribution is methodological and empirical: an evaluation framework that treats multilingual fairness as a first-class object of measurement rather than an after-the-fact slice of an English-centric benchmark, and a characterization of how current schemes behave under it.  \n- Framework. We propose four evaluation components designe","cbCail3YFc16Tkg1","https://ap.wps.com/l/cbCail3YFc16Tkg1","pdf",772120,24,"English","# Introduction\n## Motivation and problem statement\n## Proposed work and contributions\n# Related Work\n## Watermarking schemes\n## Watermark evaluation","[{\"question\":\"Why is cross-lingual fairness a concern in language model watermarking evaluations?\",\"answer\":\"Most watermark evaluations rely on English prompts and English text with scheme-specific detection thresholds. In multilingual settings, evaluation-design choices that seem irrelevant in English can substantially change conclusions across languages.\"},{\"question\":\"What is the proposed evaluation framework for multilingual watermarking?\",\"answer\":\"It combines empirically calibrated detection thresholds, a threshold-independent companion metric to separate calibration failures from detection failures, three distinct quality measurement paradigms, and a generalized-entropy decomposition of cross-language disparity over a typological partition.\"},{\"question\":\"What do the experiments reveal about where fairness disparities originate?\",\"answer\":\"Disparity is typically large, scheme-specific, and predominantly between-family on the typological partition. This suggests watermarking cross-lingual fairness gaps are structural to language properties rather than quirks of particular languages.\"}]","Auditing Cross-Lingual Fairness in Language Model Watermarking | PDF"]