{"id":"https://openalex.org/W4417447550","doi":"https://doi.org/10.18653/v1/2026.acl-long.992","title":"Rethinking Jailbreak Detection of Large Vision Language Models with Representational Contrastive Scoring","display_name":"Rethinking Jailbreak Detection of Large Vision Language Models with Representational Contrastive Scoring","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W4417447550","doi":"https://doi.org/10.18653/v1/2026.acl-long.992"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.acl-long.992","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.992","pdf_url":"https://aclanthology.org/2026.acl-long.992.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":"Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2026.acl-long.992.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":null,"display_name":"Peichun Hua","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Peichun Hua","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5005996023","display_name":"Hao Li","orcid":"https://orcid.org/0000-0003-2464-2422"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hao Li","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5011992463","display_name":"Shanghao Shi","orcid":"https://orcid.org/0009-0007-6964-8250"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shanghao Shi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5086399597","display_name":"Zhiyuan Yu","orcid":"https://orcid.org/0000-0001-6196-7598"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhiyuan Yu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":null,"display_name":"Ning Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ning Zhang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"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":0.03742143,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"21748","last_page":"21785"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.8299000263214111,"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"}},"topics":[{"id":"https://openalex.org/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.8299000263214111,"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/T11512","display_name":"Anomaly Detection Techniques and Applications","score":0.053599998354911804,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.015300000086426735,"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/generalization","display_name":"Generalization","score":0.6721000075340271},{"id":"https://openalex.org/keywords/code","display_name":"Code (set theory)","score":0.5005999803543091},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.49079999327659607},{"id":"https://openalex.org/keywords/protocol","display_name":"Protocol (science)","score":0.4535999894142151},{"id":"https://openalex.org/keywords/path","display_name":"Path (computing)","score":0.45159998536109924},{"id":"https://openalex.org/keywords/safer","display_name":"SAFER","score":0.43549999594688416},{"id":"https://openalex.org/keywords/simple","display_name":"Simple (philosophy)","score":0.38199999928474426},{"id":"https://openalex.org/keywords/projection","display_name":"Projection (relational algebra)","score":0.3601999878883362}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8126999735832214},{"id":"https://openalex.org/C177148314","wikidata":"https://www.wikidata.org/wiki/Q170084","display_name":"Generalization","level":2,"score":0.6721000075340271},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5408999919891357},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.5005999803543091},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.49079999327659607},{"id":"https://openalex.org/C2780385302","wikidata":"https://www.wikidata.org/wiki/Q367158","display_name":"Protocol (science)","level":3,"score":0.4535999894142151},{"id":"https://openalex.org/C2777735758","wikidata":"https://www.wikidata.org/wiki/Q817765","display_name":"Path (computing)","level":2,"score":0.45159998536109924},{"id":"https://openalex.org/C2776654903","wikidata":"https://www.wikidata.org/wiki/Q2601463","display_name":"SAFER","level":2,"score":0.43549999594688416},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.40529999136924744},{"id":"https://openalex.org/C2780586882","wikidata":"https://www.wikidata.org/wiki/Q7520643","display_name":"Simple (philosophy)","level":2,"score":0.38199999928474426},{"id":"https://openalex.org/C57493831","wikidata":"https://www.wikidata.org/wiki/Q3134666","display_name":"Projection (relational algebra)","level":2,"score":0.3601999878883362},{"id":"https://openalex.org/C114289077","wikidata":"https://www.wikidata.org/wiki/Q3284399","display_name":"Statistical model","level":2,"score":0.3240000009536743},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.30309998989105225},{"id":"https://openalex.org/C168167062","wikidata":"https://www.wikidata.org/wiki/Q1117970","display_name":"Component (thermodynamics)","level":2,"score":0.2928999960422516},{"id":"https://openalex.org/C66024118","wikidata":"https://www.wikidata.org/wiki/Q1122506","display_name":"Computational model","level":2,"score":0.28369998931884766},{"id":"https://openalex.org/C22367795","wikidata":"https://www.wikidata.org/wiki/Q7625208","display_name":"Structured prediction","level":2,"score":0.27790001034736633},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.2687000036239624},{"id":"https://openalex.org/C140547941","wikidata":"https://www.wikidata.org/wiki/Q7797194","display_name":"Threat model","level":2,"score":0.26420000195503235},{"id":"https://openalex.org/C195324797","wikidata":"https://www.wikidata.org/wiki/Q33742","display_name":"Natural language","level":2,"score":0.2632000148296356},{"id":"https://openalex.org/C157657479","wikidata":"https://www.wikidata.org/wiki/Q2367247","display_name":"Closed captioning","level":3,"score":0.2540999948978424},{"id":"https://openalex.org/C3017944768","wikidata":"https://www.wikidata.org/wiki/Q1450463","display_name":"Poison control","level":2,"score":0.25130000710487366}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.18653/v1/2026.acl-long.992","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.992","pdf_url":"https://aclanthology.org/2026.acl-long.992.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":"Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2512.12069","is_oa":true,"landing_page_url":"https://arxiv.org/abs/2512.12069","pdf_url":"https://arxiv.org/pdf/2512.12069","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,"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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2512.12069","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2512.12069","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,"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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.acl-long.992","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.992","pdf_url":"https://aclanthology.org/2026.acl-long.992.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":"Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G2515678443","display_name":"CPS: MEDIUM: Certified Robust Learning for Multi-Agent Planning and Control","funder_award_id":"2403758","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G6820632708","display_name":"Collaborative Research: SaTC: CORE: Medium: A Networking Perspective of Blockchain Security: Modeling, Analysis, and Defense","funder_award_id":"2154930","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G7452299184","display_name":null,"funder_award_id":"W911NF","funder_id":"https://openalex.org/F4320338281","funder_display_name":"Army Research Office"},{"id":"https://openalex.org/G8985129795","display_name":null,"funder_award_id":"W911NF-24-1-0155","funder_id":"https://openalex.org/F4320338281","funder_display_name":"Army Research Office"}],"funders":[{"id":"https://openalex.org/F4320306076","display_name":"National Science Foundation","ror":"https://ror.org/021nxhr62"},{"id":"https://openalex.org/F4320337345","display_name":"Office of Naval Research","ror":"https://ror.org/00rk2pe57"},{"id":"https://openalex.org/F4320338281","display_name":"Army Research Office","ror":"https://ror.org/05epdh915"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4417447550.pdf","grobid_xml":"https://content.openalex.org/works/W4417447550.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Large":[0],"Vision-Language":[1],"Models":[2],"(LVLMs)":[3],"are":[4,17],"vulnerable":[5],"to":[6,20,60,69,111,156,159,178],"a":[7,49,80,84,108,122,151,183],"growing":[8],"array":[9],"of":[10,104],"multimodal":[11],"jailbreak":[12,167],"attacks,":[13],"necessitating":[14],"defenses":[15],"that":[16,54,128,165],"both":[18],"generalizable":[19],"novel":[21],"threats":[22],"and":[23,115,142],"efficient":[24],"for":[25],"practical":[26,184],"deployment.Many":[27],"current":[28],"strategies":[29],"fall":[30],"short,":[31],"either":[32],"targeting":[33],"specific":[34],"attack":[35,161],"patterns,":[36],"which":[37],"limits":[38],"generalization,":[39],"or":[40],"imposing":[41],"high":[42],"computational":[43],"overhead.While":[44],"lightweight":[45,109],"anomaly-detection":[46],"methods":[47,177],"offer":[48],"promising":[50],"direction,":[51],"we":[52,74],"find":[53],"their":[55],"common":[56],"one-class":[57],"design":[58],"tends":[59],"confuse":[61],"unseen":[62,160],"benign":[63,114],"inputs":[64,117],"with":[65],"malicious":[66,116,131],"ones,":[67],"leading":[68],"unreliable":[70],"over-rejection.To":[71],"address":[72],"this,":[73],"propose":[75],"Representational":[76],"Contrastive":[77,140,145],"Scoring":[78],"(RCS),":[79],"framework":[81],"built":[82],"on":[83,150,193],"key":[85],"insight:":[86],"the":[87,94,101,179],"most":[88],"potent":[89],"safety":[90],"signals":[91],"reside":[92],"within":[93],"LVLM's":[95],"own":[96],"internal":[97,102,180],"representations.Our":[98],"approach":[99],"inspects":[100],"geometry":[103],"these":[105],"representations,":[106,181],"learning":[107],"projection":[110],"maximally":[112],"separate":[113],"in":[118],"safety-critical":[119],"layers.This":[120],"enables":[121],"simple":[123],"yet":[124],"powerful":[125],"contrastive":[126],"score":[127],"differentiates":[129],"true":[130],"intent":[132],"from":[133],"mere":[134],"distribution":[135],"shift.Our":[136],"instantiations,":[137],"MCD":[138],"(Mahalanobis":[139],"Detection)":[141],"KCD":[143],"(K-nearest":[144],"Detection),":[146],"achieve":[147],"state-of-the-art":[148],"performance":[149],"challenging":[152],"evaluation":[153],"protocol":[154],"designed":[155],"test":[157],"generalization":[158],"types.This":[162],"work":[163],"demonstrates":[164],"effective":[166],"detection":[168],"can":[169],"be":[170],"achieved":[171],"by":[172],"applying":[173],"simple,":[174],"interpretable":[175],"statistical":[176],"offering":[182],"path":[185],"towards":[186],"safer":[187],"LVLM":[188],"deployment.Our":[189],"code":[190],"is":[191],"available":[192],"Github":[194],"1":[195],".":[196]},"counts_by_year":[],"updated_date":"2026-08-05T07:39:15.569665","created_date":"2025-12-17T00:00:00"}