{"id":"https://openalex.org/W7140090786","doi":"https://doi.org/10.18653/v1/2026.findings-eacl.154","title":"Defeating Cerberus: Privacy-Leakage Mitigation in Vision Language Models","display_name":"Defeating Cerberus: Privacy-Leakage Mitigation in Vision Language Models","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7140090786","doi":"https://doi.org/10.18653/v1/2026.findings-eacl.154"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.findings-eacl.154","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-eacl.154","pdf_url":"https://aclanthology.org/2026.findings-eacl.154.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: EACL 2026","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2026.findings-eacl.154.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5146851937","display_name":"Boyang Zhang","orcid":null},"institutions":[{"id":"https://openalex.org/I4210128801","display_name":"Helmholtz Center for Information Security","ror":"https://ror.org/02njgxr09","country_code":"DE","type":"facility","lineage":["https://openalex.org/I1305996414","https://openalex.org/I4210128801"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Boyang Zhang","raw_affiliation_strings":["CISPA Helmholtz Center for Information Security ,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"CISPA Helmholtz Center for Information Security ,","institution_ids":["https://openalex.org/I4210128801"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5090968712","display_name":"\u0130stemi Ekin Akku\u015f","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Istemi Ekin Akkus","raw_affiliation_strings":["Nokia Bell Labs"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Nokia Bell Labs","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5146782484","display_name":"Ruichuan Chen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ruichuan Chen","raw_affiliation_strings":["Nokia Bell Labs"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Nokia Bell Labs","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5127777044","display_name":"Alice Dethise","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Alice Dethise","raw_affiliation_strings":["Nokia Bell Labs"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Nokia Bell Labs","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5000874049","display_name":"K. Satzke","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Klaus Satzke","raw_affiliation_strings":["Nokia Bell Labs"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Nokia Bell Labs","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5005955313","display_name":"Ivica Rimac","orcid":"https://orcid.org/0000-0003-3234-2841"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ivica Rimac","raw_affiliation_strings":["Nokia Bell Labs"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Nokia Bell Labs","institution_ids":[]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5146808741","display_name":"Yang Zhang","orcid":null},"institutions":[{"id":"https://openalex.org/I4210128801","display_name":"Helmholtz Center for Information Security","ror":"https://ror.org/02njgxr09","country_code":"DE","type":"facility","lineage":["https://openalex.org/I1305996414","https://openalex.org/I4210128801"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Yang Zhang","raw_affiliation_strings":["CISPA Helmholtz Center for Information Security ,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"CISPA Helmholtz Center for Information Security ,","institution_ids":["https://openalex.org/I4210128801"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"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":0.240025,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"2952","last_page":"2965"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.0706000030040741,"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/T10028","display_name":"Topic Modeling","score":0.0706000030040741,"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/T11045","display_name":"Privacy, Security, and Data Protection","score":0.04749999940395355,"subfield":{"id":"https://openalex.org/subfields/3312","display_name":"Sociology and Political Science"},"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/T12262","display_name":"Hate Speech and Cyberbullying Detection","score":0.04670000076293945,"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/language-model","display_name":"Language model","score":0.3815000057220459},{"id":"https://openalex.org/keywords/natural-language","display_name":"Natural language","score":0.33820000290870667},{"id":"https://openalex.org/keywords/modeling-language","display_name":"Modeling language","score":0.27559998631477356},{"id":"https://openalex.org/keywords/feature","display_name":"Feature (linguistics)","score":0.27160000801086426},{"id":"https://openalex.org/keywords/data-modeling","display_name":"Data modeling","score":0.2526000142097473}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6621999740600586},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4569999873638153},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.41040000319480896},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.3815000057220459},{"id":"https://openalex.org/C195324797","wikidata":"https://www.wikidata.org/wiki/Q33742","display_name":"Natural language","level":2,"score":0.33820000290870667},{"id":"https://openalex.org/C41895202","wikidata":"https://www.wikidata.org/wiki/Q8162","display_name":"Linguistics","level":1,"score":0.31310001015663147},{"id":"https://openalex.org/C179603123","wikidata":"https://www.wikidata.org/wiki/Q1941921","display_name":"Modeling language","level":3,"score":0.27559998631477356},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.27160000801086426},{"id":"https://openalex.org/C67186912","wikidata":"https://www.wikidata.org/wiki/Q367664","display_name":"Data modeling","level":2,"score":0.2526000142097473},{"id":"https://openalex.org/C2983448237","wikidata":"https://www.wikidata.org/wiki/Q1078276","display_name":"Language understanding","level":2,"score":0.24320000410079956}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2026.findings-eacl.154","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-eacl.154","pdf_url":"https://aclanthology.org/2026.findings-eacl.154.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: EACL 2026","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.findings-eacl.154","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-eacl.154","pdf_url":"https://aclanthology.org/2026.findings-eacl.154.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: EACL 2026","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7140090786.pdf","grobid_xml":"https://content.openalex.org/works/W7140090786.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Vision":[0],"Language":[1],"Models":[2],"(VLMs)":[3],"have":[4,46],"demonstrated":[5],"remarkable":[6],"capabilities":[7],"in":[8,42],"processing":[9],"multimodal":[10,44,96],"data,":[11],"but":[12],"their":[13],"advanced":[14],"abilities":[15],"also":[16,90],"raise":[17],"significant":[18],"privacy":[19],"concerns,":[20],"particularly":[21],"regarding":[22],"Personally":[23],"Identifiable":[24],"Information":[25],"(PII)":[26],"leakage.While":[27],"relevant":[28],"research":[29],"has":[30],"been":[31],"conducted":[32],"on":[33,113,125],"single-modal":[34],"language":[35],"models":[36],"to":[37,48,78,137],"some":[38],"extent,":[39],"the":[40,43,66,92,109,131,139],"vulnerabilities":[41],"setting":[45],"yet":[47],"be":[49],"fully":[50],"investigated.Our":[51],"work":[52],"assesses":[53],"these":[54],"emerging":[55],"risks":[56],"and":[57,64,83],"introduces":[58],"a":[59],"concept-guided":[60],"mitigation":[61],"approach.By":[62],"identifying":[63],"modifying":[65],"model's":[67],"internal":[68],"states":[69],"associated":[70],"with":[71,122],"PIIrelated":[72],"content,":[73],"our":[74,142],"method":[75,110],"guides":[76],"VLMs":[77],"refuse":[79],"PII-sensitive":[80],"tasks":[81,121],"effectively":[82],"efficiently,":[84],"without":[85],"requiring":[86],"re-training":[87],"or":[88],"finetuning.We":[89],"address":[91],"current":[93],"lack":[94],"of":[95,141],"PII":[97],"datasets":[98],"by":[99],"constructing":[100],"various":[101,119,135],"ones":[102],"that":[103],"simulate":[104],"real-world":[105],"scenarios.Experimental":[106],"results":[107],"demonstrate":[108],"can":[111],"achieve":[112],"average":[114],"93.3%":[115],"refusal":[116],"rate":[117],"for":[118],"PII-related":[120],"minimal":[123],"impact":[124],"unrelated":[126],"model":[127],"performances.We":[128],"further":[129],"examine":[130],"mitigation's":[132],"performance":[133],"under":[134],"conditions":[136],"show":[138],"adaptability":[140],"proposed":[143],"method.":[144]},"counts_by_year":[],"updated_date":"2026-08-18T07:49:30.821534","created_date":"2026-03-04T00:00:00"}