{"id":"https://openalex.org/W4412888793","doi":"https://doi.org/10.18653/v1/2025.findings-acl.105","title":"SafeRoute: Adaptive Model Selection for Efficient and Accurate Safety Guardrails in Large Language Models","display_name":"SafeRoute: Adaptive Model Selection for Efficient and Accurate Safety Guardrails in Large Language Models","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4412888793","doi":"https://doi.org/10.18653/v1/2025.findings-acl.105"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.findings-acl.105","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.105","pdf_url":"https://aclanthology.org/2025.findings-acl.105.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: ACL 2025","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.findings-acl.105.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5000293382","display_name":"Seanie Lee","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Seanie Lee","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5021139954","display_name":"Dong Bok Lee","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Dong Bok Lee","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Dominik Wagner","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Dominik Wagner","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Minki Kang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Minki Kang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Haebin Seong","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Haebin Seong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5033302750","display_name":"Tobias Bocklet","orcid":"https://orcid.org/0009-0008-7780-8821"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tobias Bocklet","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5064342381","display_name":"Ju Ho Lee","orcid":"https://orcid.org/0000-0002-1642-5922"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Juho Lee","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5070302452","display_name":"Sung Ju Hwang","orcid":"https://orcid.org/0000-0002-9675-2324"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sung Ju Hwang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":8,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.7368,"has_fulltext":true,"cited_by_count":1,"citation_normalized_percentile":{"value":0.72427779,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":95,"max":98},"biblio":{"volume":null,"issue":null,"first_page":"2053","last_page":"2069"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11344","display_name":"Traffic Prediction and Management Techniques","score":0.9681000113487244,"subfield":{"id":"https://openalex.org/subfields/2215","display_name":"Building and Construction"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T11344","display_name":"Traffic Prediction and Management Techniques","score":0.9681000113487244,"subfield":{"id":"https://openalex.org/subfields/2215","display_name":"Building and Construction"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10761","display_name":"Vehicular Ad Hoc Networks (VANETs)","score":0.9370999932289124,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T11606","display_name":"Infrastructure Maintenance and Monitoring","score":0.9074000120162964,"subfield":{"id":"https://openalex.org/subfields/2205","display_name":"Civil and Structural Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7343398332595825},{"id":"https://openalex.org/keywords/selection","display_name":"Selection (genetic algorithm)","score":0.6388561725616455},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.20356261730194092}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7343398332595825},{"id":"https://openalex.org/C81917197","wikidata":"https://www.wikidata.org/wiki/Q628760","display_name":"Selection (genetic algorithm)","level":2,"score":0.6388561725616455},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.20356261730194092}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.findings-acl.105","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.105","pdf_url":"https://aclanthology.org/2025.findings-acl.105.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: ACL 2025","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.findings-acl.105","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.105","pdf_url":"https://aclanthology.org/2025.findings-acl.105.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: ACL 2025","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G3736587842","display_name":null,"funder_award_id":"RS-2019-II190075","funder_id":"https://openalex.org/F4320335489","funder_display_name":"Institute for Information and Communications Technology Promotion"},{"id":"https://openalex.org/G4468082379","display_name":null,"funder_award_id":"RS-2019-II190075","funder_id":"https://openalex.org/F4320328359","funder_display_name":"Ministry of Science and ICT, South Korea"},{"id":"https://openalex.org/G4700831490","display_name":null,"funder_award_id":"2022-","funder_id":"https://openalex.org/F4320335489","funder_display_name":"Institute for Information and Communications Technology Promotion"}],"funders":[{"id":"https://openalex.org/F4320320671","display_name":"National Research Foundation","ror":"https://ror.org/05s0g1g46"},{"id":"https://openalex.org/F4320322120","display_name":"National Research Foundation of Korea","ror":"https://ror.org/013aysd81"},{"id":"https://openalex.org/F4320324161","display_name":"Korea Advanced Institute of Science and Technology","ror":"https://ror.org/05apxxy63"},{"id":"https://openalex.org/F4320328359","display_name":"Ministry of Science and ICT, South Korea","ror":"https://ror.org/01wpjm123"},{"id":"https://openalex.org/F4320332195","display_name":"Samsung","ror":"https://ror.org/04w3jy968"},{"id":"https://openalex.org/F4320335489","display_name":"Institute for Information and Communications Technology Promotion","ror":"https://ror.org/01g0hqq23"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4412888793.pdf","grobid_xml":"https://content.openalex.org/works/W4412888793.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W2390279801","https://openalex.org/W4391913857","https://openalex.org/W2358668433","https://openalex.org/W4396701345","https://openalex.org/W2376932109","https://openalex.org/W2001405890","https://openalex.org/W4396696052"],"abstract_inverted_index":{"Deploying":[0],"large":[1,20],"language":[2,149],"models":[3,12,23,36],"(LLMs)":[4],"in":[5],"real-world":[6],"applications":[7],"requires":[8],"robust":[9],"safety":[10,21,95,117,139],"guard":[11,22,96,118],"to":[13,98,112],"detect":[14],"and":[15,138],"block":[16],"harmful":[17,148],"user":[18],"prompts.While":[19],"achieve":[24],"strong":[25],"performance,":[26,140],"their":[27],"computational":[28,136],"cost":[29,137],"is":[30],"substantial.To":[31],"mitigate":[32],"this,":[33,76],"smaller":[34,63],"distilled":[35],"are":[37],"used,":[38],"but":[39],"they":[40],"often":[41],"underperform":[42],"on":[43,121],"\"hard\"":[44],"examples":[45,86],"where":[46],"the":[47,62,71,93,99,102,115,133],"larger":[48,72,94,116],"model":[49,97,129,150],"provides":[50],"accurate":[51],"predictions.We":[52],"observe":[53],"that":[54,83,101,126],"many":[55],"inputs":[56],"can":[57],"be":[58],"reliably":[59],"handled":[60],"by":[61,75],"model,":[64],"while":[65,108],"only":[66],"a":[67,80],"small":[68],"fraction":[69],"require":[70],"model's":[73],"capacity.Motivated":[74],"we":[77],"propose":[78],"SafeRoute,":[79],"binary":[81],"router":[82,103],"distinguishes":[84],"hard":[85],"from":[87],"easy":[88],"ones.Our":[89],"method":[90],"selectively":[91],"applies":[92],"data":[100],"considers":[104],"hard,":[105],"improving":[106],"efficiency":[107],"maintaining":[109],"accuracy":[110],"compared":[111],"solely":[113],"using":[114],"model.Experimental":[119],"results":[120],"multiple":[122],"benchmark":[123],"datasets":[124],"demonstrate":[125],"our":[127],"adaptive":[128],"selection":[130],"significantly":[131],"enhances":[132],"trade-off":[134],"between":[135],"outperforming":[141],"relevant":[142],"baselines.Warning:":[143],"This":[144],"paper":[145],"contains":[146],"potentially":[147],"outputs.":[151]},"counts_by_year":[{"year":2026,"cited_by_count":1}],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2025-10-10T00:00:00"}