{"id":"https://openalex.org/W7166876835","doi":"https://doi.org/10.18653/v1/2026.acl-long.72","title":"Leave My Images Alone: Preventing Multi-Modal Large Language Models from Analyzing Images via Visual Prompt Injection","display_name":"Leave My Images Alone: Preventing Multi-Modal Large Language Models from Analyzing Images via Visual Prompt Injection","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7166876835","doi":"https://doi.org/10.18653/v1/2026.acl-long.72"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.acl-long.72","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.72","pdf_url":"https://aclanthology.org/2026.acl-long.72.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":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2026.acl-long.72.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5111345581","display_name":"Zedian Shao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zedian Shao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139775856","display_name":"Hongbin Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hongbin Liu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139826758","display_name":"Yuepeng Hu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yuepeng Hu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5139820860","display_name":"Neil Zhenqiang Gong","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Neil Zhenqiang Gong","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":0.88209272,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"1588","last_page":"1604"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":0.7172999978065491,"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"}},"topics":[{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":0.7172999978065491,"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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.09030000120401382,"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/T10799","display_name":"Data Visualization and Analytics","score":0.02889999933540821,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/visualization","display_name":"Visualization","score":0.38370001316070557},{"id":"https://openalex.org/keywords/feature","display_name":"Feature (linguistics)","score":0.3255999982357025},{"id":"https://openalex.org/keywords/visual-language","display_name":"Visual language","score":0.28110000491142273},{"id":"https://openalex.org/keywords/set","display_name":"Set (abstract data type)","score":0.2425999939441681},{"id":"https://openalex.org/keywords/natural-language","display_name":"Natural language","score":0.24210000038146973}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5873000025749207},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5544999837875366},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.5246999859809875},{"id":"https://openalex.org/C36464697","wikidata":"https://www.wikidata.org/wiki/Q451553","display_name":"Visualization","level":2,"score":0.38370001316070557},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.3255999982357025},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.2865999937057495},{"id":"https://openalex.org/C2780878386","wikidata":"https://www.wikidata.org/wiki/Q1659648","display_name":"Visual language","level":2,"score":0.28110000491142273},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.2425999939441681},{"id":"https://openalex.org/C195324797","wikidata":"https://www.wikidata.org/wiki/Q33742","display_name":"Natural language","level":2,"score":0.24210000038146973},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.24130000174045563}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2026.acl-long.72","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.72","pdf_url":"https://aclanthology.org/2026.acl-long.72.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"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.acl-long.72","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.72","pdf_url":"https://aclanthology.org/2026.acl-long.72.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/G652313190","display_name":"Collaborative Research: SaTC: CORE: Small: Securing Recommender Systems against Data Poisoning Attacks","funder_award_id":"2125977","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G8377316191","display_name":"AI Institute for Edge Computing Leveraging Next Generation Networks (Athena)","funder_award_id":"2112562","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G8991437584","display_name":"Collaborative Research: SaTC: CORE: Medium: Securing LLMs against Prompt Injection Attacks","funder_award_id":"2450935","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":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7166876835.pdf","grobid_xml":"https://content.openalex.org/works/W7166876835.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Multi-modal":[0],"large":[1],"language":[2],"models":[3],"(MLLMs)":[4],"have":[5],"emerged":[6],"as":[7,42,73,104],"powerful":[8],"tools":[9],"for":[10],"analyzing":[11],"Internet-scale":[12],"image":[13,89,167],"data,":[14],"offering":[15],"significant":[16],"benefits":[17],"but":[18],"also":[19],"raising":[20],"critical":[21],"safety":[22],"and":[23,122,133,136,164,169,174],"societal":[24],"concerns.In":[25],"particular,":[26],"open-weight":[27,162],"MLLMs":[28,121,163],"may":[29],"be":[30],"misused":[31],"to":[32,98],"extract":[33],"sensitive":[34],"information":[35],"from":[36],"personal":[37],"images":[38,60],"at":[39],"scale,":[40],"such":[41,103],"identities,":[43],"locations,":[44],"or":[45],"other":[46],"private":[47],"details.In":[48],"this":[49],"work,":[50],"we":[51,125],"propose":[52],"ImageProtector,":[53,146],"a":[54,65,74,81,87,100],"user-side":[55],"method":[56],"that":[57,71,111,138],"proactively":[58],"protects":[59],"before":[61],"sharing":[62],"by":[63],"embedding":[64],"carefully":[66],"crafted,":[67],"nearly":[68],"imperceptible":[69],"perturbation":[70],"acts":[72],"visual":[75],"prompt":[76],"injection":[77],"attack":[78],"on":[79,156],"MLLMs.As":[80],"result,":[82],"when":[83],"an":[84,91],"adversary":[85],"analyzes":[86],"protected":[88],"with":[90,110],"MLLM,":[92],"the":[93,115,143,157,172,175],"MLLM":[94],"is":[95],"consistently":[96],"induced":[97],"generate":[99],"refusal":[101],"response":[102],"\"I'm":[105],"sorry,":[106],"I":[107],"can't":[108],"help":[109],"request.\"We":[112],"empirically":[113],"demonstrate":[114],"effectiveness":[116],"of":[117,145,161,177],"ImageProtector":[118],"across":[119],"six":[120],"four":[123],"datasets.Additionally,":[124],"evaluate":[126],"three":[127],"potential":[128],"countermeasures,":[129],"Gaussian":[130],"noise,":[131],"DiffPure,":[132],"adversarial":[134],"training,":[135],"show":[137],"while":[139],"they":[140,147],"partially":[141],"mitigate":[142],"impact":[144],"simultaneously":[148],"degrade":[149],"model":[150],"accuracy":[151],"and/or":[152],"efficiency.Our":[153],"study":[154],"focuses":[155],"practically":[158],"important":[159],"setting":[160],"large-scale":[165],"automated":[166],"analysis,":[168],"highlights":[170],"both":[171],"promise":[173],"limitations":[176],"perturbation-based":[178],"privacy":[179],"protection.":[180]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2026-07-02T00:00:00"}