{"id":"https://openalex.org/W7140093486","doi":"https://doi.org/10.18653/v1/2026.eacl-long.197","title":"Enhancing the Safety of Medical Vision-Language Models by Synthetic Demonstrations","display_name":"Enhancing the Safety of Medical Vision-Language Models by Synthetic Demonstrations","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7140093486","doi":"https://doi.org/10.18653/v1/2026.eacl-long.197"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2026.eacl-long.197","doi":"https://doi.org/10.18653/v1/2026.eacl-long.197","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.eacl-long.197","pdf_url":"https://aclanthology.org/2026.eacl-long.197.pdf","source":{"id":"https://openalex.org/S4306418011","display_name":"Conference of the European Chapter of the Association for Computational Linguistics","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"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 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"volume":null,"type":"conference-paper","indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"hybrid","oa_url":"https://aclanthology.org/2026.eacl-long.197.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5040852802","display_name":"Zhiyu Xue","orcid":"https://orcid.org/0009-0009-1389-0499","observed_orcids":["https://orcid.org/0009-0009-1389-0499"]},"institutions":[{"id":"https://openalex.org/I154570441","display_name":"University of California, Santa Barbara","ror":"https://ror.org/02t274463","country_code":"US","type":"education","lineage":["https://openalex.org/I154570441"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Zhiyu Xue","raw_affiliation_strings":["UC Santa Barbara ,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"UC Santa Barbara ,","institution_ids":["https://openalex.org/I154570441"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5029714016","display_name":"Reza Abbasi-Asl","orcid":"https://orcid.org/0000-0001-7824-4628","observed_orcids":["https://orcid.org/0000-0001-7824-4628"]},"institutions":[{"id":"https://openalex.org/I180670191","display_name":"University of California, San Francisco","ror":"https://ror.org/043mz5j54","country_code":"US","type":"education","lineage":["https://openalex.org/I180670191"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Reza Abbasi-Asl","raw_affiliation_strings":["UC San Francisco"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"UC San Francisco","institution_ids":["https://openalex.org/I180670191"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5040270189","display_name":"Ramtin Pedarsani","orcid":"https://orcid.org/0000-0002-1126-0292","observed_orcids":["https://orcid.org/0000-0002-1126-0292","https://orcid.org/0000-0003-0786-2132"]},"institutions":[{"id":"https://openalex.org/I154570441","display_name":"University of California, Santa Barbara","ror":"https://ror.org/02t274463","country_code":"US","type":"education","lineage":["https://openalex.org/I154570441"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Ramtin Pedarsani","raw_affiliation_strings":["UC Santa Barbara ,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"UC Santa Barbara ,","institution_ids":["https://openalex.org/I154570441"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"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":0.14633552,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"4207","last_page":"4220"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.9984999895095825,"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":0.9984999895095825,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.00039999998989515007,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/T11424","display_name":"Security and Verification in Computing","score":0.00019999999494757503,"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/medical-vision-language-models","display_name":"medical vision-language models","score":0.9620000123977661},{"id":"https://openalex.org/keywords/model-safety","display_name":"model safety","score":0.6729999780654907},{"id":"https://openalex.org/keywords/jailbreak-attacks","display_name":"jailbreak attacks","score":0.8650000095367432},{"id":"https://openalex.org/keywords/security-performance-trade-off","display_name":"security-performance trade-off","score":0.7260000109672546},{"id":"https://openalex.org/keywords/diagnostic-imaging","display_name":"diagnostic imaging","score":0.41100001335144043}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.48570001125335693},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.3052000105381012},{"id":"https://openalex.org/C112930515","wikidata":"https://www.wikidata.org/wiki/Q4389547","display_name":"Risk analysis (engineering)","level":1,"score":0.2896000146865845},{"id":"https://openalex.org/C201995342","wikidata":"https://www.wikidata.org/wiki/Q682496","display_name":"Systems engineering","level":1,"score":0.28459998965263367},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.24740000069141388},{"id":"https://openalex.org/C18762648","wikidata":"https://www.wikidata.org/wiki/Q42213","display_name":"Work (physics)","level":2,"score":0.24629999697208405},{"id":"https://openalex.org/C2779530757","wikidata":"https://www.wikidata.org/wiki/Q1207505","display_name":"Quality (philosophy)","level":2,"score":0.23890000581741333},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.23720000684261322},{"id":"https://openalex.org/C168167062","wikidata":"https://www.wikidata.org/wiki/Q1117970","display_name":"Component (thermodynamics)","level":2,"score":0.23409999907016754},{"id":"https://openalex.org/C2779662365","wikidata":"https://www.wikidata.org/wiki/Q5416694","display_name":"Event (particle physics)","level":2,"score":0.22059999406337738}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.18653/v1/2026.eacl-long.197","doi":"https://doi.org/10.18653/v1/2026.eacl-long.197","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.eacl-long.197","pdf_url":"https://aclanthology.org/2026.eacl-long.197.pdf","source":{"id":"https://openalex.org/S4306418011","display_name":"Conference of the European Chapter of the Association for Computational Linguistics","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"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 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2506.09067","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2506.09067","pdf_url":"https://arxiv.org/pdf/2506.09067","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,"listed_in":[],"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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2506.09067","doi":"https://doi.org/10.48550/arxiv.2506.09067","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2506.09067","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,"listed_in":[],"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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.eacl-long.197","doi":"https://doi.org/10.18653/v1/2026.eacl-long.197","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.eacl-long.197","pdf_url":"https://aclanthology.org/2026.eacl-long.197.pdf","source":{"id":"https://openalex.org/S4306418011","display_name":"Conference of the European Chapter of the Association for Computational Linguistics","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"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 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"sustainable_development_goals":[{"score":0.504800021648407,"id":"https://metadata.un.org/sdg/3","display_name":"Good health and well-being"}],"sustainable_development_goals_aurora":[],"study_designs":[],"awards":[{"id":"https://openalex.org/G2081463616","display_name":"Collaborative Research: CIF: Small: Robust Machine Learning under Sparse Adversarial Attacks","funder_award_id":"2236483","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G3662856300","display_name":"Fair Acquisition of Private Data in Multi-agent Systems","funder_award_id":"2419982","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G3876709226","display_name":"NSF-NSERC: Fairness Fundamentals: Geometry-inspired Algorithms and Long-term Implications","funder_award_id":"2342253","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"}],"funders":[{"id":"https://openalex.org/F4320306076","display_name":"National Science Foundation","ror":"https://ror.org/021nxhr62"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7140093486.pdf","grobid_xml":"https://content.openalex.org/works/W7140093486.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4416844981","https://openalex.org/W7117102973","https://openalex.org/W4417277192","https://openalex.org/W4407023155","https://openalex.org/W4413639576","https://openalex.org/W7148462735","https://openalex.org/W4407683663","https://openalex.org/W4403345328","https://openalex.org/W4401201064","https://openalex.org/W7140309924"],"abstract_inverted_index":{"Generative":[0],"medical":[1,23,103],"vision-language":[2],"models":[3],"(Med-VLMs)":[4],"are":[5],"primarily":[6],"designed":[7],"to":[8,76,90],"generate":[9],"complex":[10],"textual":[11,99],"information":[12],"(e.g.,":[13,22,28],"diagnostic":[14],"reports)":[15],"from":[16,107],"multimodal":[17],"inputs":[18],"including":[19],"vision":[20],"modality":[21,27],"images)":[24],"and":[25,98,152],"language":[26],"clinical":[29,79,119],"queries).However,":[30],"their":[31],"security":[32,60,151],"vulnerabilities":[33],"remain":[34],"underexplored.Med-VLMs":[35],"should":[36],"be":[37],"capable":[38],"of":[39,65],"rejecting":[40],"harmful":[41,92],"queries,":[42,93],"such":[43],"as":[44,145],"Provide":[45],"detailed":[46],"instructions":[47],"for":[48,53,149],"using":[49],"this":[50,81],"CT":[51],"scan":[52],"insurance":[54],"fraud.At":[55],"the":[56,63,132,136],"same":[57],"time,":[58],"addressing":[59],"concerns":[61],"introduces":[62],"risk":[64],"over-defense,":[66],"where":[67],"safetyenhancing":[68],"mechanisms":[69],"may":[70],"degrade":[71],"general":[72],"performance,":[73],"causing":[74],"Med-VLMs":[75],"reject":[77],"benign":[78],"queries.In":[80],"paper,":[82],"we":[83,110,128],"propose":[84],"a":[85,141,146],"novel":[86],"inference-time":[87],"defense":[88,95,114],"strategy":[89,115,144],"mitigate":[91],"enabling":[94],"against":[96],"visual":[97],"jailbreak":[100],"attacks.Using":[101],"diverse":[102],"imaging":[104],"datasets":[105],"collected":[106],"nine":[108],"modalities,":[109],"demonstrate":[111],"that":[112,130],"our":[113],"based":[116],"on":[117],"synthetic":[118],"demonstrations":[120],"enhances":[121],"model":[122],"safety":[123],"without":[124],"significantly":[125],"compromising":[126],"performance.Additionally,":[127],"find":[129],"increasing":[131],"demonstration":[133,143,156],"budget":[134,157],"alleviates":[135],"overdefense":[137],"issue.We":[138],"then":[139],"introduce":[140],"mixed":[142],"trade-off":[147],"solution":[148],"balancing":[150],"performance":[153],"under":[154],"few-shot":[155],"constraints":[158],"1":[159],".":[160]},"counts_by_year":[],"updated_date":"2026-10-12T09:40:04.860316","created_date":"2026-03-24T00:00:00"}