{"id":"https://openalex.org/W7166818064","doi":"https://doi.org/10.18653/v1/2026.findings-acl.983","title":"Hard to Read, Easy to Jailbreak: How Visual Degradation Bypasses MLLM Safety Alignment","display_name":"Hard to Read, Easy to Jailbreak: How Visual Degradation Bypasses MLLM Safety Alignment","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7166818064","doi":"https://doi.org/10.18653/v1/2026.findings-acl.983"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.findings-acl.983","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.983","pdf_url":"https://aclanthology.org/2026.findings-acl.983.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: ACL 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-acl.983.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5139782408","display_name":"Zhixue Song","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhixue Song","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5089249830","display_name":"Boyan Han","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Boyan Han","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139804582","display_name":"Yiwei Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yiwei Wang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5139749506","display_name":"Chi Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chi 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":0.85151007,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"19643","last_page":"19658"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10914","display_name":"Tactile and Sensory Interactions","score":0.07500000298023224,"subfield":{"id":"https://openalex.org/subfields/2805","display_name":"Cognitive Neuroscience"},"field":{"id":"https://openalex.org/fields/28","display_name":"Neuroscience"},"domain":{"id":"https://openalex.org/domains/1","display_name":"Life Sciences"}},"topics":[{"id":"https://openalex.org/T10914","display_name":"Tactile and Sensory Interactions","score":0.07500000298023224,"subfield":{"id":"https://openalex.org/subfields/2805","display_name":"Cognitive Neuroscience"},"field":{"id":"https://openalex.org/fields/28","display_name":"Neuroscience"},"domain":{"id":"https://openalex.org/domains/1","display_name":"Life Sciences"}},{"id":"https://openalex.org/T10918","display_name":"Memory Processes and Influences","score":0.05310000106692314,"subfield":{"id":"https://openalex.org/subfields/2805","display_name":"Cognitive Neuroscience"},"field":{"id":"https://openalex.org/fields/28","display_name":"Neuroscience"},"domain":{"id":"https://openalex.org/domains/1","display_name":"Life Sciences"}},{"id":"https://openalex.org/T14056","display_name":"Safety Warnings and Signage","score":0.04190000146627426,"subfield":{"id":"https://openalex.org/subfields/3207","display_name":"Social Psychology"},"field":{"id":"https://openalex.org/fields/32","display_name":"Psychology"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/degradation","display_name":"Degradation (telecommunications)","score":0.4569000005722046},{"id":"https://openalex.org/keywords/work","display_name":"Work (physics)","score":0.2734000086784363},{"id":"https://openalex.org/keywords/visualization","display_name":"Visualization","score":0.258899986743927}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5288000106811523},{"id":"https://openalex.org/C2779679103","wikidata":"https://www.wikidata.org/wiki/Q5251805","display_name":"Degradation (telecommunications)","level":2,"score":0.4569000005722046},{"id":"https://openalex.org/C199639397","wikidata":"https://www.wikidata.org/wiki/Q1788588","display_name":"Engineering drawing","level":1,"score":0.30979999899864197},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.30959999561309814},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.2994999885559082},{"id":"https://openalex.org/C192562407","wikidata":"https://www.wikidata.org/wiki/Q228736","display_name":"Materials science","level":0,"score":0.29100000858306885},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.2768000066280365},{"id":"https://openalex.org/C18762648","wikidata":"https://www.wikidata.org/wiki/Q42213","display_name":"Work (physics)","level":2,"score":0.2734000086784363},{"id":"https://openalex.org/C36464697","wikidata":"https://www.wikidata.org/wiki/Q451553","display_name":"Visualization","level":2,"score":0.258899986743927},{"id":"https://openalex.org/C21880701","wikidata":"https://www.wikidata.org/wiki/Q2144042","display_name":"Process engineering","level":1,"score":0.25119999051094055}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2026.findings-acl.983","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.983","pdf_url":"https://aclanthology.org/2026.findings-acl.983.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: ACL 2026","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.findings-acl.983","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.983","pdf_url":"https://aclanthology.org/2026.findings-acl.983.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: ACL 2026","raw_type":"proceedings-article"},"sustainable_development_goals":[{"display_name":"Climate action","score":0.5881754159927368,"id":"https://metadata.un.org/sdg/13"}],"awards":[{"id":"https://openalex.org/G5961903033","display_name":null,"funder_award_id":"6250070674","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"}],"funders":[{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7166818064.pdf","grobid_xml":"https://content.openalex.org/works/W7166818064.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Recent":[0],"advancements":[1],"in":[2,117],"visual":[3,79,107],"context":[4],"compression":[5,119],"enable":[6],"MLLMs":[7],"to":[8,24,56,64,105],"process":[9],"ultra-long":[10],"contexts":[11],"efficiently":[12],"by":[13,100],"rendering":[14],"text":[15,51],"into":[16],"images.However,":[17],"we":[18,88],"identify":[19],"a":[20,90,102,114],"critical":[21,122],"vulnerability":[22],"inherent":[23],"this":[25,55],"paradigm:":[26],"lowering":[27],"image":[28],"resolution":[29,45],"inadvertently":[30],"catalyzes":[31],"jailbreaking.Our":[32],"experiments":[33],"reveal":[34],"that":[35,60,96],"the":[36,61,125],"safety":[37,72,110],"defenses":[38],"of":[39,128],"SOTA":[40],"models":[41],"deteriorate":[42],"sharply":[43],"as":[44],"degrades,":[46],"surprisingly":[47],"persisting":[48],"even":[49],"when":[50],"remains":[52],"legible.We":[53],"attribute":[54],"\"Cognitive":[57],"Overload\",":[58],"hypothesizing":[59],"effort":[62],"required":[63],"decipher":[65],"degraded":[66],"inputs":[67],"diverts":[68],"attentional":[69],"resources":[70],"from":[71,109],"auditing.This":[73],"phenomenon":[74],"is":[75],"consistent":[76],"across":[77],"various":[78],"perturbations,":[80],"including":[81],"noise":[82],"and":[83,120],"geometric":[84],"distortion.To":[85],"address":[86],"this,":[87],"propose":[89],"simple":[91],"\"Structured":[92],"Cognitive":[93],"Offloading\"":[94],"strategy":[95],"mitigates":[97],"these":[98],"risks":[99],"enforcing":[101],"serialized":[103],"pipeline":[104],"decouple":[106],"transcription":[108],"assessment.Our":[111],"work":[112],"exposes":[113],"significant":[115],"risk":[116],"vision-based":[118],"provides":[121],"insights":[123],"for":[124],"secure":[126],"design":[127],"future":[129],"MLLMs.":[130]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2026-07-02T00:00:00"}