{"id":"https://openalex.org/W4412886794","doi":"https://doi.org/10.18653/v1/2025.acl-long.146","title":"Con Instruction: Universal Jailbreaking of Multimodal Large Language Models via Non-Textual Modalities","display_name":"Con Instruction: Universal Jailbreaking of Multimodal Large Language Models via Non-Textual Modalities","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4412886794","doi":"https://doi.org/10.18653/v1/2025.acl-long.146"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.acl-long.146","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.146","pdf_url":"https://aclanthology.org/2025.acl-long.146.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 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.acl-long.146.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5073054282","display_name":"Jiahui Geng","orcid":"https://orcid.org/0000-0002-4205-8230"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jiahui Geng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5061574554","display_name":"Thy Thy Tran","orcid":"https://orcid.org/0000-0002-0627-9706"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Thy Thy Tran","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5012055259","display_name":"Preslav Nakov","orcid":"https://orcid.org/0000-0002-3600-1510"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Preslav Nakov","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5027450194","display_name":"Iryna Gurevych","orcid":"https://orcid.org/0000-0003-2187-7621"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Iryna Gurevych","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":3.3008,"has_fulltext":true,"cited_by_count":2,"citation_normalized_percentile":{"value":0.92077425,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":91,"max":97},"biblio":{"volume":null,"issue":null,"first_page":"2917","last_page":"2933"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.9779000282287598,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.9779000282287598,"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/T10028","display_name":"Topic Modeling","score":0.9732999801635742,"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/T13643","display_name":"Artificial Intelligence in Law","score":0.9391000270843506,"subfield":{"id":"https://openalex.org/subfields/3320","display_name":"Political Science and International Relations"},"field":{"id":"https://openalex.org/fields/33","display_name":"Social Sciences"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7738754153251648},{"id":"https://openalex.org/keywords/modalities","display_name":"Modalities","score":0.7491637468338013},{"id":"https://openalex.org/keywords/natural-language-processing","display_name":"Natural language processing","score":0.5524970293045044},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.43758291006088257},{"id":"https://openalex.org/keywords/linguistics","display_name":"Linguistics","score":0.35701656341552734},{"id":"https://openalex.org/keywords/sociology","display_name":"Sociology","score":0.05446624755859375},{"id":"https://openalex.org/keywords/philosophy","display_name":"Philosophy","score":0.04979822039604187}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7738754153251648},{"id":"https://openalex.org/C2779903281","wikidata":"https://www.wikidata.org/wiki/Q6888026","display_name":"Modalities","level":2,"score":0.7491637468338013},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.5524970293045044},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.43758291006088257},{"id":"https://openalex.org/C41895202","wikidata":"https://www.wikidata.org/wiki/Q8162","display_name":"Linguistics","level":1,"score":0.35701656341552734},{"id":"https://openalex.org/C144024400","wikidata":"https://www.wikidata.org/wiki/Q21201","display_name":"Sociology","level":0,"score":0.05446624755859375},{"id":"https://openalex.org/C138885662","wikidata":"https://www.wikidata.org/wiki/Q5891","display_name":"Philosophy","level":0,"score":0.04979822039604187},{"id":"https://openalex.org/C36289849","wikidata":"https://www.wikidata.org/wiki/Q34749","display_name":"Social science","level":1,"score":0.0}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.18653/v1/2025.acl-long.146","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.146","pdf_url":"https://aclanthology.org/2025.acl-long.146.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 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},{"id":"pmh:oai:tubiblio.ulb.tu-darmstadt.de:155156","is_oa":false,"landing_page_url":"http://tubiblio.ulb.tu-darmstadt.de/view/person/Geng=3AJiahui=3A=3A.html>","pdf_url":null,"source":{"id":"https://openalex.org/S4377196390","display_name":"TUbilio (Technical University of Darmstadt)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I31512782","host_organization_name":"Technische Universit\u00e4t Darmstadt","host_organization_lineage":["https://openalex.org/I31512782"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"acceptedVersion","is_accepted":true,"is_published":false,"raw_source_name":null,"raw_type":"Konferenzver\u00f6ffentlichung"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.acl-long.146","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.146","pdf_url":"https://aclanthology.org/2025.acl-long.146.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 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/4","display_name":"Quality Education","score":0.4699999988079071}],"awards":[],"funders":[{"id":"https://openalex.org/F4320320300","display_name":"European Commission","ror":"https://ror.org/00k4n6c32"},{"id":"https://openalex.org/F4320334322","display_name":"HORIZON EUROPE Framework Programme","ror":null}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4412886794.pdf","grobid_xml":"https://content.openalex.org/works/W4412886794.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W2185469136","https://openalex.org/W2011264131","https://openalex.org/W2390279801","https://openalex.org/W4391913857","https://openalex.org/W2358668433","https://openalex.org/W4306353150","https://openalex.org/W3204019825"],"abstract_inverted_index":{"Existing":[0],"attacks":[1,170],"against":[2,168],"multimodal":[3],"language":[4],"models":[5],"(MLLMs)":[6],"primarily":[7],"communicate":[8],"instructions":[9,47,113],"through":[10],"text":[11,89],"accompanied":[12],"by":[13,32],"adversarial":[14,28,40,77],"images.In":[15],"contrast,":[16],"here":[17],"we":[18,164],"exploit":[19],"the":[20,39,49,53,111,125,149,161],"capabilities":[21],"of":[22,56,72],"MLLMs":[23,82],"to":[24,42,110,114],"interpret":[25],"non-textual":[26,76],"instructions-specifically":[27],"images":[29],"or":[30,70],"audio-generated":[31],"our":[33,63,146,169],"novel":[34],"method,":[35],"Con":[36,121],"Instruction.We":[37],"optimize":[38],"examples":[41,78],"align":[43],"closely":[44],"with":[45,87],"target":[46],"in":[48,59,128],"embedding":[50],"space,":[51],"revealing":[52],"detrimental":[54],"aspects":[55],"sophisticated":[57],"understanding":[58],"MLLMs.Unlike":[60],"previous":[61],"work,":[62],"method":[64,147],"does":[65],"not":[66],"require":[67],"training":[68],"data":[69],"preprocessing":[71],"textual":[73],"instructions.While":[74],"these":[75],"can":[79],"effectively":[80,123],"bypass":[81],"safety":[83,126],"mechanisms,":[84],"their":[85],"combination":[86],"various":[88,129,166],"inputs":[90],"substantially":[91],"amplifies":[92],"attack":[93,99,116,151],"success.We":[94],"further":[95],"introduce":[96],"a":[97,173],"new":[98],"response":[100,106],"categorization":[101],"(ARC)":[102],"that":[103,120],"considers":[104],"both":[105],"quality":[107],"and":[108,131,137,144,156,171],"relevance":[109],"malicious":[112],"evaluate":[115],"success.The":[117],"results":[118],"show":[119],"Instruction":[122],"bypasses":[124],"mechanisms":[127],"visual":[130],"audio-language":[132],"models,":[133],"including":[134],"LLaVA-v1.5,InternVL,":[135],"Qwen-VL,":[136],"Qwen-Audio,":[138],"across":[139],"two":[140],"standard":[141],"benchmarks:":[142],"AdvBench":[143],"SafeBench.Specifically,":[145],"achieves":[148],"highest":[150],"success":[152],"rates,":[153],"reaching":[154],"81.3%":[155],"86.6%":[157],"on":[158],"LLaVA-v1.5":[159],"(13B).On":[160],"defense":[162],"side,":[163],"explore":[165],"methods":[167],"find":[172],"substantial":[174],"gap":[175],"among":[176],"existing":[177],"techniques.Our":[178],"implementation":[179],"is":[180],"made":[181],"available.":[182]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2025,"cited_by_count":1}],"updated_date":"2026-08-04T08:18:43.703281","created_date":"2025-10-10T00:00:00"}