{"id":"https://openalex.org/W4413847135","doi":"https://doi.org/10.1109/tai.2025.3603547","title":"Online Safety Analysis for LLMs: A Benchmark, an Assessment, and a Path Forward","display_name":"Online Safety Analysis for LLMs: A Benchmark, an Assessment, and a Path Forward","publication_year":2025,"publication_date":"2025-08-29","ids":{"openalex":"https://openalex.org/W4413847135","doi":"https://doi.org/10.1109/tai.2025.3603547"},"language":"en","primary_location":{"id":"doi:10.1109/tai.2025.3603547","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tai.2025.3603547","pdf_url":null,"source":{"id":"https://openalex.org/S4210169448","display_name":"IEEE Transactions on Artificial Intelligence","issn_l":"2691-4581","issn":["2691-4581"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Artificial Intelligence","raw_type":"journal-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":false,"oa_status":"closed","oa_url":null,"any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5034067129","display_name":"Xuan Xie","orcid":"https://orcid.org/0000-0003-3981-8515"},"institutions":[{"id":"https://openalex.org/I111950717","display_name":"Macau University of Science and Technology","ror":"https://ror.org/03jqs2n27","country_code":"MO","type":"education","lineage":["https://openalex.org/I111950717","https://openalex.org/I4391767947"]}],"countries":["MO"],"is_corresponding":true,"raw_author_name":"Xuan Xie","raw_affiliation_strings":["School of Computer Science and Engineering, Macau University of Science and Technology, Macau, China"],"raw_orcid":"https://orcid.org/0000-0003-3981-8515","affiliations":[{"raw_affiliation_string":"School of Computer Science and Engineering, Macau University of Science and Technology, Macau, China","institution_ids":["https://openalex.org/I111950717"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5104291087","display_name":"Jiayang Song","orcid":"https://orcid.org/0009-0008-7093-9781"},"institutions":[{"id":"https://openalex.org/I111950717","display_name":"Macau University of Science and Technology","ror":"https://ror.org/03jqs2n27","country_code":"MO","type":"education","lineage":["https://openalex.org/I111950717","https://openalex.org/I4391767947"]}],"countries":["MO"],"is_corresponding":false,"raw_author_name":"Jiayang Song","raw_affiliation_strings":["School of Computer Science and Engineering, Macau University of Science and Technology, Macau, China"],"raw_orcid":"https://orcid.org/0009-0008-7093-9781","affiliations":[{"raw_affiliation_string":"School of Computer Science and Engineering, Macau University of Science and Technology, Macau, China","institution_ids":["https://openalex.org/I111950717"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5026971084","display_name":"Zhehua Zhou","orcid":"https://orcid.org/0000-0001-9542-4858"},"institutions":[{"id":"https://openalex.org/I204512498","display_name":"University of Macau","ror":"https://ror.org/01r4q9n85","country_code":"MO","type":"education","lineage":["https://openalex.org/I204512498"]}],"countries":["MO"],"is_corresponding":false,"raw_author_name":"Zhehua Zhou","raw_affiliation_strings":["University of Macau, Macau, China"],"raw_orcid":"https://orcid.org/0000-0001-9542-4858","affiliations":[{"raw_affiliation_string":"University of Macau, Macau, China","institution_ids":["https://openalex.org/I204512498"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101684275","display_name":"Yuheng Huang","orcid":"https://orcid.org/0000-0003-3666-4020"},"institutions":[{"id":"https://openalex.org/I74801974","display_name":"The University of Tokyo","ror":"https://ror.org/057zh3y96","country_code":"JP","type":"education","lineage":["https://openalex.org/I74801974"]}],"countries":["JP"],"is_corresponding":false,"raw_author_name":"Yuheng Huang","raw_affiliation_strings":["The University of Tokyo, Tokyo, Japan","University of Tokyo, Japan"],"raw_orcid":"https://orcid.org/0000-0003-3666-4020","affiliations":[{"raw_affiliation_string":"The University of Tokyo, Tokyo, Japan","institution_ids":["https://openalex.org/I74801974"]},{"raw_affiliation_string":"University of Tokyo, Japan","institution_ids":["https://openalex.org/I74801974"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Da Song","orcid":"https://orcid.org/0000-0001-9267-4229"},"institutions":[{"id":"https://openalex.org/I4210164802","display_name":"Mila - Quebec Artificial Intelligence Institute","ror":"https://ror.org/05c22rx21","country_code":"CA","type":"facility","lineage":["https://openalex.org/I4210164802"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Da Song","raw_affiliation_strings":["Mila - Québec AI Institute, Montréal, QC, Canada","Mila - Québec AI Institutey, Canada"],"raw_orcid":"https://orcid.org/0000-0001-9267-4229","affiliations":[{"raw_affiliation_string":"Mila - Québec AI Institute, Montréal, QC, Canada","institution_ids":["https://openalex.org/I4210164802"]},{"raw_affiliation_string":"Mila - Québec AI Institutey, Canada","institution_ids":["https://openalex.org/I4210164802"]}]},{"author_position":"last","author":{"id":null,"display_name":"Lei Ma","orcid":"https://orcid.org/0000-0002-8621-2420"},"institutions":[{"id":"https://openalex.org/I74801974","display_name":"The University of Tokyo","ror":"https://ror.org/057zh3y96","country_code":"JP","type":"education","lineage":["https://openalex.org/I74801974"]}],"countries":["JP"],"is_corresponding":false,"raw_author_name":"Lei Ma","raw_affiliation_strings":["The University of Tokyo, Tokyo, Japan","The University of Tokyo, Japan"],"raw_orcid":"https://orcid.org/0000-0002-8621-2420","affiliations":[{"raw_affiliation_string":"The University of Tokyo, Tokyo, Japan","institution_ids":["https://openalex.org/I74801974"]},{"raw_affiliation_string":"The University of Tokyo, Japan","institution_ids":["https://openalex.org/I74801974"]}]}],"institutions":[],"countries_distinct_count":3,"institutions_distinct_count":6,"corresponding_author_ids":["https://openalex.org/A5034067129"],"corresponding_institution_ids":["https://openalex.org/I111950717"],"apc_list":null,"apc_paid":null,"fwci":5.8102,"has_fulltext":false,"cited_by_count":2,"citation_normalized_percentile":{"value":0.96233543,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":95,"max":96},"biblio":{"volume":"7","issue":"3","first_page":"1626","last_page":"1641"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T13999","display_name":"Digital Rights Management and Security","score":0.7746999859809875,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T13999","display_name":"Digital Rights Management and Security","score":0.7746999859809875,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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/T13295","display_name":"Safety Systems Engineering in Autonomy","score":0.7164999842643738,"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/benchmark","display_name":"Benchmark (surveying)","score":0.6895914673805237},{"id":"https://openalex.org/keywords/path","display_name":"Path (computing)","score":0.5748812556266785},{"id":"https://openalex.org/keywords/path-analysis","display_name":"Path analysis (statistics)","score":0.4377690553665161},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.38749659061431885},{"id":"https://openalex.org/keywords/risk-analysis","display_name":"Risk analysis (engineering)","score":0.33549606800079346},{"id":"https://openalex.org/keywords/business","display_name":"Business","score":0.2600257992744446},{"id":"https://openalex.org/keywords/geography","display_name":"Geography","score":0.17438238859176636},{"id":"https://openalex.org/keywords/machine-learning","display_name":"Machine learning","score":0.14318349957466125},{"id":"https://openalex.org/keywords/computer-network","display_name":"Computer network","score":0.08883500099182129},{"id":"https://openalex.org/keywords/cartography","display_name":"Cartography","score":0.08434441685676575}],"concepts":[{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.6895914673805237},{"id":"https://openalex.org/C2777735758","wikidata":"https://www.wikidata.org/wiki/Q817765","display_name":"Path (computing)","level":2,"score":0.5748812556266785},{"id":"https://openalex.org/C82793941","wikidata":"https://www.wikidata.org/wiki/Q1046024","display_name":"Path analysis (statistics)","level":2,"score":0.4377690553665161},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.38749659061431885},{"id":"https://openalex.org/C112930515","wikidata":"https://www.wikidata.org/wiki/Q4389547","display_name":"Risk analysis (engineering)","level":1,"score":0.33549606800079346},{"id":"https://openalex.org/C144133560","wikidata":"https://www.wikidata.org/wiki/Q4830453","display_name":"Business","level":0,"score":0.2600257992744446},{"id":"https://openalex.org/C205649164","wikidata":"https://www.wikidata.org/wiki/Q1071","display_name":"Geography","level":0,"score":0.17438238859176636},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.14318349957466125},{"id":"https://openalex.org/C31258907","wikidata":"https://www.wikidata.org/wiki/Q1301371","display_name":"Computer network","level":1,"score":0.08883500099182129},{"id":"https://openalex.org/C58640448","wikidata":"https://www.wikidata.org/wiki/Q42515","display_name":"Cartography","level":1,"score":0.08434441685676575}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/tai.2025.3603547","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tai.2025.3603547","pdf_url":null,"source":{"id":"https://openalex.org/S4210169448","display_name":"IEEE Transactions on Artificial Intelligence","issn_l":"2691-4581","issn":["2691-4581"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Artificial Intelligence","raw_type":"journal-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W2378211422","https://openalex.org/W4321353415","https://openalex.org/W2745001401","https://openalex.org/W2130974462","https://openalex.org/W2028665553","https://openalex.org/W2086519370","https://openalex.org/W4246352526"],"abstract_inverted_index":{"While":[0],"Large":[1],"Language":[2],"Models":[3],"(LLMs)":[4],"have":[5],"seen":[6],"widespread":[7],"applications":[8],"across":[9,234],"numerous":[10],"fields,":[11],"their":[12,18,231],"limited":[13],"interpretability":[14],"poses":[15],"concerns":[16],"regarding":[17],"safe":[19],"operations":[20],"from":[21],"multiple":[22,199],"aspects,":[23],"e.g.,":[24],"truthfulness":[25],"and":[26,136,156,164,169,184],"toxicity.":[27],"Recent":[28],"research":[29],"has":[30],"started":[31],"developing":[32],"quality":[33],"assurance":[34,226],"methods":[35,90,152,168,200],"for":[36,61,125,221,228],"LLMs,":[37,126,229],"introducing":[38],"techniques":[39],"such":[40],"as":[41],"offline":[42],"detectors":[43],"or":[44],"uncertainty":[45],"estimation":[46],"methods.":[47],"However,":[48],"these":[49],"approaches":[50],"mainly":[51],"focus":[52],"on":[53,91,153,180],"post-generation":[54],"analysis,":[55],"leaving":[56],"the":[57,64,83,101,108,116,145,162,175,191,209,222],"online":[58,87,122,149,212],"safety":[59,88,123,150,205,213],"analysis":[60,89,124,151,160],"LLMs":[62],"during":[63],"generation":[65,110],"phase":[66],"an":[67],"unexplored":[68],"area.":[69],"To":[70],"bridge":[71],"this":[72,77,140],"gap,":[73],"we":[74,114,142,188],"conduct":[75],"in":[76,107],"work":[78],"a":[79,96,128,203,218],"comprehensive":[80],"evaluation":[81,137],"of":[82,85,103,121,131,147,166,193,211,224],"effectiveness":[84],"existing":[86],"LLMs.":[92,158],"We":[93],"begin":[94],"with":[95],"pilot":[97],"study":[98],"that":[99],"validates":[100],"feasibility":[102],"detecting":[104],"unsafe":[105],"outputs":[106],"early":[109],"process.":[111],"Following":[112],"this,":[113],"establish":[115],"first":[117],"publicly":[118],"available":[119],"benchmark":[120],"including":[127],"broad":[129],"spectrum":[130],"methods,":[132,196],"models,":[133],"tasks,":[134],"datasets,":[135],"metrics.":[138],"Utilizing":[139],"benchmark,":[141],"extensively":[143],"analyze":[144],"performance":[146],"state-of-the-art":[148],"both":[154],"open-source":[155],"closed-source":[157],"This":[159],"reveals":[161],"strengths":[163],"weaknesses":[165],"individual":[167],"offers":[170],"valuable":[171],"insights":[172],"into":[173],"selecting":[174],"most":[176],"appropriate":[177],"method":[178],"based":[179],"specific":[181],"application":[182],"scenarios":[183],"task":[185],"requirements.":[186],"Furthermore,":[187],"also":[189],"explore":[190],"potential":[192],"using":[194],"hybridization":[195],"i.e.,":[197],"combining":[198],"to":[201,207],"derive":[202],"collective":[204],"conclusion,":[206],"enhance":[208],"efficacy":[210],"analysis.":[214],"Our":[215],"findings":[216],"indicate":[217],"promising":[219],"direction":[220],"development":[223],"trustworthy":[225],"methodologies":[227],"facilitating":[230],"reliable":[232],"deployments":[233],"diverse":[235],"domains.":[236]},"counts_by_year":[{"year":2025,"cited_by_count":2}],"updated_date":"2026-03-01T06:05:34.837733","created_date":"2025-10-10T00:00:00"}