{"id":"https://openalex.org/W4416034379","doi":"https://doi.org/10.18653/v1/2025.findings-emnlp.1193","title":"Safety Through Reasoning: An Empirical Study of Reasoning Guardrail Models","display_name":"Safety Through Reasoning: An Empirical Study of Reasoning Guardrail Models","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4416034379","doi":"https://doi.org/10.18653/v1/2025.findings-emnlp.1193"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2025.findings-emnlp.1193","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.1193","pdf_url":"https://aclanthology.org/2025.findings-emnlp.1193.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: EMNLP 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-emnlp.1193.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5111090196","display_name":"Makesh Narsimhan Sreedhar","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Makesh Narsimhan Sreedhar","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5077365712","display_name":"Traian Rebedea","orcid":"https://orcid.org/0000-0002-7255-5537"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Traian Rebedea","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5001562064","display_name":"Christopher Parisien","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Christopher Parisien","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":3,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.9617,"has_fulltext":true,"cited_by_count":1,"citation_normalized_percentile":{"value":0.79646835,"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":"21862","last_page":"21880"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T13295","display_name":"Safety Systems Engineering in Autonomy","score":0.14820000529289246,"subfield":{"id":"https://openalex.org/subfields/2213","display_name":"Safety, Risk, Reliability and Quality"},"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/T13295","display_name":"Safety Systems Engineering in Autonomy","score":0.14820000529289246,"subfield":{"id":"https://openalex.org/subfields/2213","display_name":"Safety, Risk, Reliability and Quality"},"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/T11099","display_name":"Autonomous Vehicle Technology and Safety","score":0.1356000006198883,"subfield":{"id":"https://openalex.org/subfields/2203","display_name":"Automotive 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/T10370","display_name":"Traffic and Road Safety","score":0.06040000170469284,"subfield":{"id":"https://openalex.org/subfields/2213","display_name":"Safety, Risk, Reliability and Quality"},"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/empirical-research","display_name":"Empirical research","score":0.5924000144004822},{"id":"https://openalex.org/keywords/accident","display_name":"Accident (philosophy)","score":0.28439998626708984},{"id":"https://openalex.org/keywords/empirical-evidence","display_name":"Empirical evidence","score":0.2508000135421753},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.2395000010728836}],"concepts":[{"id":"https://openalex.org/C120936955","wikidata":"https://www.wikidata.org/wiki/Q2155640","display_name":"Empirical research","level":2,"score":0.5924000144004822},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5034999847412109},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3077999949455261},{"id":"https://openalex.org/C2780289543","wikidata":"https://www.wikidata.org/wiki/Q424630","display_name":"Accident (philosophy)","level":2,"score":0.28439998626708984},{"id":"https://openalex.org/C149782125","wikidata":"https://www.wikidata.org/wiki/Q160039","display_name":"Econometrics","level":1,"score":0.2822999954223633},{"id":"https://openalex.org/C166052673","wikidata":"https://www.wikidata.org/wiki/Q83021","display_name":"Empirical evidence","level":2,"score":0.2508000135421753},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.24140000343322754},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.2395000010728836},{"id":"https://openalex.org/C42475967","wikidata":"https://www.wikidata.org/wiki/Q194292","display_name":"Operations research","level":1,"score":0.22769999504089355},{"id":"https://openalex.org/C94966114","wikidata":"https://www.wikidata.org/wiki/Q29256","display_name":"Black box","level":2,"score":0.2232999950647354}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.findings-emnlp.1193","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.1193","pdf_url":"https://aclanthology.org/2025.findings-emnlp.1193.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: EMNLP 2025","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.findings-emnlp.1193","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.1193","pdf_url":"https://aclanthology.org/2025.findings-emnlp.1193.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: EMNLP 2025","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4416034379.pdf","grobid_xml":"https://content.openalex.org/works/W4416034379.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Reasoning-based":[0],"language":[1],"models":[2,47,81,166],"have":[3],"demonstrated":[4],"strong":[5,83],"performance":[6,88],"across":[7],"various":[8],"domains,":[9],"with":[10,51,89],"the":[11,74,99,103,116,128],"most":[12],"notable":[13],"gains":[14],"seen":[15],"in":[16,167],"mathematical":[17],"and":[18,33,71,135,137,155,159,162],"coding":[19],"tasks.Recent":[20],"research":[21],"has":[22],"shown":[23],"that":[24,79,111],"reasoning":[25,125,131,146],"also":[26],"offers":[27],"significant":[28],"benefits":[29],"for":[30,48,106,153],"LLM":[31],"safety":[32,58],"guardrail":[34,46],"applications.In":[35],"this":[36],"work,":[37],"we":[38,77,119],"conduct":[39],"a":[40],"comprehensive":[41],"analysis":[42],"of":[43,130],"training":[44,92,140],"reasoning-based":[45,80,164],"content":[49],"moderation,":[50],"an":[52],"emphasis":[53],"on":[54,65,133],"generalization":[55],"to":[56,101,141,157],"custom":[57],"policies":[59],"at":[60],"inference":[61,72,117],"time.Our":[62],"study":[63],"focuses":[64],"two":[66],"key":[67],"dimensions:":[68],"data":[69,75,105],"efficiency":[70],"efficiency.On":[73],"front,":[76],"find":[78],"exhibit":[82],"sample":[84],"efficiency,":[85],"achieving":[86],"competitive":[87],"significantly":[90],"fewer":[91],"examples":[93],"than":[94],"their":[95],"non-reasoning":[96],"counterparts.This":[97],"unlocks":[98],"potential":[100],"repurpose":[102],"remaining":[104],"mining":[107],"high-value,":[108],"difficult":[109],"samples":[110],"further":[112],"enhance":[113],"model":[114],"performance.On":[115],"side,":[118],"evaluate":[120],"practical":[121,151],"trade-offs":[122],"by":[123],"introducing":[124],"budgets,":[126],"examining":[127],"impact":[129],"length":[132],"latency":[134],"accuracy,":[136],"exploring":[138],"dual-mode":[139],"allow":[142],"runtime":[143],"control":[144],"over":[145],"behavior.Our":[147],"findings":[148],"will":[149],"provide":[150],"insights":[152],"researchers":[154],"developers":[156],"effectively":[158],"efficiently":[160],"train":[161],"deploy":[163],"guardrails":[165],"real-world":[168],"systems.":[169]},"counts_by_year":[{"year":2026,"cited_by_count":1}],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2025-11-08T00:00:00"}