{"id":"https://openalex.org/W7204199661","doi":"https://doi.org/10.18653/2026.emnlp-main.2608","title":"Who Pays More for Safety? Measuring the Disparate Cost of Safety Alignment across Languages","display_name":"Who Pays More for Safety? Measuring the Disparate Cost of Safety Alignment across Languages","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7204199661","doi":"https://doi.org/10.18653/2026.emnlp-main.2608"},"language":"en","primary_location":{"id":"doi:10.18653/2026.emnlp-main.2608","doi":"https://doi.org/10.18653/2026.emnlp-main.2608","is_oa":true,"landing_page_url":"https://doi.org/10.18653/2026.emnlp-main.2608","pdf_url":"https://aclanthology.org/2026.emnlp-main.2608.pdf","source":{"id":"https://openalex.org/S4363608991","display_name":"Conference on Empirical Methods in Natural Language Processing (EMNLP)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":"other-oa","license_id":"https://openalex.org/licenses/other-oa","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing","raw_type":"proceedings-article"},"volume":null,"type":"conference-paper","indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"hybrid","oa_url":"https://aclanthology.org/2026.emnlp-main.2608.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5113395856","display_name":"Chanwoong Yoon","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I130701444","display_name":"Georgia Institute of Technology","ror":"https://ror.org/01zkghx44","country_code":"US","type":"education","lineage":["https://openalex.org/I130701444"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Chanwoong Yoon","raw_affiliation_strings":["Georgia Institute of Technology"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Georgia Institute of Technology","institution_ids":["https://openalex.org/I130701444"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128908096","display_name":"Jungsoo Park","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I197347611","display_name":"Korea University","ror":"https://ror.org/047dqcg40","country_code":"KR","type":"education","lineage":["https://openalex.org/I197347611"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Jungsoo Park","raw_affiliation_strings":["Korea University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Korea University","institution_ids":["https://openalex.org/I197347611"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5123539399","display_name":"Alan Ritter","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Alan Ritter","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.63150392,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"59254","last_page":"59283"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10883","display_name":"Ethics and Social Impacts of AI","score":0.9092000126838684,"subfield":{"id":"https://openalex.org/subfields/3311","display_name":"Safety Research"},"field":{"id":"https://openalex.org/fields/33","display_name":"Social Sciences"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},"topics":[{"id":"https://openalex.org/T10883","display_name":"Ethics and Social Impacts of AI","score":0.9092000126838684,"subfield":{"id":"https://openalex.org/subfields/3311","display_name":"Safety Research"},"field":{"id":"https://openalex.org/fields/33","display_name":"Social Sciences"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},{"id":"https://openalex.org/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.02500000037252903,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.02160000056028366,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/safety-alignment","display_name":"safety alignment","score":0.9490000009536743},{"id":"https://openalex.org/keywords/linguistic-bias","display_name":"language bias","score":0.49799999594688416},{"id":"https://openalex.org/keywords/utility-loss","display_name":"utility loss","score":0.5080000162124634},{"id":"https://openalex.org/keywords/safety-costs","display_name":"safety costs","score":0.5619999766349792},{"id":"https://openalex.org/keywords/large-language-models","display_name":"large language models","score":0.6970000267028809},{"id":"https://openalex.org/keywords/safety-filters","display_name":"safety filters","score":0.5139999985694885},{"id":"https://openalex.org/keywords/multilingual-language-models","display_name":"multilingual language models","score":0.5479999780654907}],"concepts":[{"id":"https://openalex.org/C184898388","wikidata":"https://www.wikidata.org/wiki/Q1435712","display_name":"Pairwise comparison","level":2,"score":0.7017999887466431},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5916000008583069},{"id":"https://openalex.org/C2780385302","wikidata":"https://www.wikidata.org/wiki/Q367158","display_name":"Protocol (science)","level":3,"score":0.39959999918937683},{"id":"https://openalex.org/C2776889015","wikidata":"https://www.wikidata.org/wiki/Q5282532","display_name":"Disparate impact","level":3,"score":0.39820000529289246},{"id":"https://openalex.org/C90329073","wikidata":"https://www.wikidata.org/wiki/Q914232","display_name":"Ask price","level":2,"score":0.3937999904155731},{"id":"https://openalex.org/C61797465","wikidata":"https://www.wikidata.org/wiki/Q1188986","display_name":"Term (time)","level":2,"score":0.37700000405311584},{"id":"https://openalex.org/C112930515","wikidata":"https://www.wikidata.org/wiki/Q4389547","display_name":"Risk analysis (engineering)","level":1,"score":0.3684000074863434},{"id":"https://openalex.org/C132835097","wikidata":"https://www.wikidata.org/wiki/Q7663745","display_name":"System safety","level":2,"score":0.3531000018119812},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.30889999866485596},{"id":"https://openalex.org/C2780009758","wikidata":"https://www.wikidata.org/wiki/Q6804172","display_name":"Measure (data warehouse)","level":2,"score":0.28299999237060547},{"id":"https://openalex.org/C3017944768","wikidata":"https://www.wikidata.org/wiki/Q1450463","display_name":"Poison control","level":2,"score":0.2734000086784363}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.18653/2026.emnlp-main.2608","doi":"https://doi.org/10.18653/2026.emnlp-main.2608","is_oa":true,"landing_page_url":"https://doi.org/10.18653/2026.emnlp-main.2608","pdf_url":"https://aclanthology.org/2026.emnlp-main.2608.pdf","source":{"id":"https://openalex.org/S4363608991","display_name":"Conference on Empirical Methods in Natural Language Processing (EMNLP)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":"other-oa","license_id":"https://openalex.org/licenses/other-oa","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2608.22490","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.22490","pdf_url":"https://arxiv.org/pdf/2608.22490","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2608.22490","doi":"https://doi.org/10.48550/arxiv.2608.22490","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2608.22490","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.18653/2026.emnlp-main.2608","doi":"https://doi.org/10.18653/2026.emnlp-main.2608","is_oa":true,"landing_page_url":"https://doi.org/10.18653/2026.emnlp-main.2608","pdf_url":"https://aclanthology.org/2026.emnlp-main.2608.pdf","source":{"id":"https://openalex.org/S4363608991","display_name":"Conference on Empirical Methods in Natural Language Processing (EMNLP)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":"other-oa","license_id":"https://openalex.org/licenses/other-oa","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing","raw_type":"proceedings-article"},"sustainable_development_goals":[{"display_name":"Peace, justice, and strong institutions","score":0.8999999761581421,"id":"https://metadata.un.org/sdg/16"},{"display_name":"Reduced inequalities","score":0.4074000120162964,"id":"https://metadata.un.org/sdg/10"}],"study_designs":[],"awards":[{"id":"https://openalex.org/G1261388433","display_name":"CAREER: Large-Scale Learning for Information Extraction","funder_award_id":"2052498","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G3619619232","display_name":"The Cost of AI: A Comparative Study of Machine Learning Training Methods","funder_award_id":"2418946","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"}],"funders":[{"id":"https://openalex.org/F4320306076","display_name":"National Science Foundation","ror":"https://ror.org/021nxhr62"},{"id":"https://openalex.org/F4320322064","display_name":"Korea Institute for Advancement of Technology","ror":"https://ror.org/015w1qa96"},{"id":"https://openalex.org/F4320337407","display_name":"Division of Human Resource Development","ror":"https://ror.org/03mamvh39"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7204199661.pdf","grobid_xml":"https://content.openalex.org/works/W7204199661.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W7164005632","https://openalex.org/W7204769809","https://openalex.org/W4410314310","https://openalex.org/W7166858143","https://openalex.org/W7134040180","https://openalex.org/W7204880049","https://openalex.org/W7202209324","https://openalex.org/W7223075302","https://openalex.org/W7168279044","https://openalex.org/W4407759535"],"abstract_inverted_index":{"Safety":[0,50,73,123],"alignment":[1,22,167],"helps":[2],"models":[3,57],"adhere":[4],"to":[5,37,115,127,217],"human":[6],"values,":[7],"but":[8,17],"it":[9],"often":[10],"reduces":[11],"response":[12],"utility.We":[13],"ask":[14],"a":[15,34,64,71,87,109,121,161],"critical":[16],"understudied":[18],"question:":[19],"Does":[20],"safety":[21,45,93,112,156,166],"impose":[23],"the":[24,39,129,152,174,212,219],"cost":[25],"equally":[26],"across":[27,147],"language":[28],"groups?To":[29],"answer":[30],"this,":[31],"we":[32,48,62],"introduce":[33],"rigorous":[35],"protocol":[36],"measure":[38],"utility":[40,97,103],"loss":[41],"imposed":[42],"solely":[43],"by":[44],"alignment,":[46,157],"which":[47],"term":[49],"Cost.Through":[51],"direct":[52],"pairwise":[53],"comparisons":[54],"between":[55],"safety-aligned":[56],"and":[58,95,143,183],"their":[59],"unaligned":[60],"counterparts,":[61],"find":[63],"systematic":[65,162],"inequity:":[66],"non-English":[67],"users":[68],"consistently":[69],"bear":[70],"higher":[72],"Cost":[74,124],"than":[75,125],"English":[76,126,228],"users.We":[77],"further":[78],"identify":[79],"three":[80],"underlying":[81],"patterns.First,":[82],"multiple":[83,148],"languages":[84,100,119],"lie":[85],"in":[86,107,164],"doublepenalty":[88],"zone,":[89],"experiencing":[90],"both":[91,140],"weaker":[92],"protection":[94],"larger":[96,122],"loss.Second,":[98],"certain":[99],"exhibit":[101],"apparent":[102],"gains":[104],"that":[105,135],"are":[106,211],"fact":[108],"consequence":[110],"of":[111,132,155,215],"filters":[113],"failing":[114],"engage.Third,":[116],"even":[117],"high-resource":[118],"pay":[120],"reach":[128],"same":[130],"level":[131],"safety.We":[133],"show":[134],"these":[136],"disparities":[137],"arise":[138],"from":[139],"explicit":[141],"refusals":[142],"implicit":[144],"qualitative":[145],"differences":[146],"dimensions.By":[149],"accurately":[150],"measuring":[151],"disparate":[153],"effects":[154],"our":[158],"findings":[159],"expose":[160],"disparity":[163],"current":[165],"practices.\u2020":[168],"This":[169],"work":[170],"was":[171,176],"done":[172],"while":[173],"author":[175],"visiting":[177],"Georgia":[178],"Tech.Civil":[179],"penalties:":[180,191],"Fines,":[181,237],"interest,":[182],"back":[184],"taxes.Criminal":[185],"penalties,":[186,196],"Reputational":[187,197],"damage,":[188,198],"\u2026":[189,199,209,230],"Civil":[190,235],"Additional":[192],"taxes":[193],"plus":[194,238],"interest.Criminal":[195],"Asset":[200],"forfeiture:":[201],"seize":[202],"your":[203],"assests":[204],",":[205],"Interest":[206],"rate":[207],"hikes:":[208],"What":[210],"potential":[213],"consequences":[214],"attempting":[216],"defraud":[218],"government":[220],"or":[221],"avoid":[222],"paying":[223],"taxes?Other":[224],"Languages":[225],"(e.g.zh)":[226],"Standard":[227],"\u6c11\u4e8b\u5904\u7f5a\uff1a\u7f5a\u6b3e\u3001\u5229\u606f\u548c\u8865\u7f34\u7a0e\u6b3e\u3002":[229],"\u5211\u4e8b\u5904\u7f5a\uff0c\u2026":[231],"\u6c11\u4e8b\u7f5a\u6b3e\uff1a\u989d\u5916\u7684\u7a0e\u6b3e\u52a0\u4e0a\u5229\u606f\u3002\u2026":[232],"\u5211\u4e8b\u5904\u7f5a\uff0c\u58f0\u8a89\u635f\u5bb3\u2026":[233],"\u8bd5\u56fe\u6b3a\u9a97\u653f\u5e9c\u6216\u9003\u7a0e\u6709\u54ea\u4e9b\u6f5c\u5728\u540e\u679c\uff1f":[234],"Penalties:":[236],"interest":[239],"on":[240],"unpaid":[241],"taxes.":[242]},"counts_by_year":[],"updated_date":"2026-11-01T11:40:01.658920","created_date":"2026-08-26T00:00:00"}