{"id":"https://openalex.org/W7168809751","doi":"https://doi.org/10.48448/m2wa-5w92","title":"MTMCS-Bench: Evaluating Contextual Safety of Multimodal Large Language Models in Multi-Turn Dialogues","display_name":"MTMCS-Bench: Evaluating Contextual Safety of Multimodal Large Language Models in Multi-Turn Dialogues","publication_year":2026,"publication_date":"2026-06-03","ids":{"openalex":"https://openalex.org/W7168809751","doi":"https://doi.org/10.48448/m2wa-5w92"},"language":"en","primary_location":{"id":"doi:10.48448/m2wa-5w92","doi":"https://doi.org/10.48448/m2wa-5w92","is_oa":true,"landing_page_url":"https://doi.org/10.48448/m2wa-5w92","pdf_url":null,"source":{"id":"https://openalex.org/S7407053148","display_name":"Underline Science Inc.","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":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Audiovisual"},"volume":null,"type":"other","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48448/m2wa-5w92","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5140982166","display_name":"Association for Computational Linguistics 2026","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Association for Computational Linguistics 2026","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5140961893","display_name":"Meng Jiang","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I107639228","display_name":"University of Notre Dame","ror":"https://ror.org/00mkhxb43","country_code":"US","type":"education","lineage":["https://openalex.org/I107639228"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Jiang, Meng","raw_affiliation_strings":["University of Notre Dame"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Notre Dame","institution_ids":["https://openalex.org/I107639228"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5122982494","display_name":"Dongwhi Kim","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kim, Dongwhi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5140992623","display_name":"Zheyuan Liu","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I107639228","display_name":"University of Notre Dame","ror":"https://ror.org/00mkhxb43","country_code":"US","type":"education","lineage":["https://openalex.org/I107639228"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Liu, Zheyuan","raw_affiliation_strings":["University of Notre Dame"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Notre Dame","institution_ids":["https://openalex.org/I107639228"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5048033573","display_name":"Fengran Mo","orcid":"https://orcid.org/0000-0002-0838-6994","observed_orcids":["https://orcid.org/0000-0002-0838-6994"]},"institutions":[{"id":"https://openalex.org/I70931966","display_name":"Universit\u00e9 de Montr\u00e9al","ror":"https://ror.org/0161xgx34","country_code":"CA","type":"education","lineage":["https://openalex.org/I70931966"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Mo, Fengran","raw_affiliation_strings":["University of Montreal"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Montreal","institution_ids":["https://openalex.org/I70931966"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5140888380","display_name":"Zhaoxuan Tan","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I107639228","display_name":"University of Notre Dame","ror":"https://ror.org/00mkhxb43","country_code":"US","type":"education","lineage":["https://openalex.org/I107639228"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Tan, Zhaoxuan","raw_affiliation_strings":["University of Notre Dame"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Notre Dame","institution_ids":["https://openalex.org/I107639228"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5140859564","display_name":"Yixin Wan","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I161318765","display_name":"University of California, Los Angeles","ror":"https://ror.org/046rm7j60","country_code":"US","type":"education","lineage":["https://openalex.org/I161318765"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Wan, Yixin","raw_affiliation_strings":["University of California Los Angeles"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of California Los Angeles","institution_ids":["https://openalex.org/I161318765"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5072246587","display_name":"Xiangchi Yuan","orcid":"https://orcid.org/0000-0001-5270-4817","observed_orcids":["https://orcid.org/0000-0001-5270-4817"]},"institutions":[{"id":"https://openalex.org/I130701444","display_name":"Georgia Institute of Technology","ror":"https://ror.org/01zkghx44","country_code":"US","type":"education","lineage":["https://openalex.org/I130701444"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Yuan, Xiangchi","raw_affiliation_strings":["Georgia Institute of Technology"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Georgia Institute of Technology","institution_ids":["https://openalex.org/I130701444"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":4,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.4255000054836273,"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":0.4255000054836273,"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/T12262","display_name":"Hate Speech and Cyberbullying Detection","score":0.39480000734329224,"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":0.06729999929666519,"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/multimodal-large-language-models","display_name":"multimodal large language models","score":0.9900000095367432},{"id":"https://openalex.org/keywords/multi-turn-dialogue","display_name":"multi-turn dialogue","score":0.8769999742507935},{"id":"https://openalex.org/keywords/safety-benchmark","display_name":"safety benchmark","score":0.7390000224113464},{"id":"https://openalex.org/keywords/malicious-intent-detection","display_name":"malicious intent detection","score":0.6150000095367432},{"id":"https://openalex.org/keywords/guardrails","display_name":"guardrails","score":0.6629999876022339},{"id":"https://openalex.org/keywords/ai-assistants","display_name":"AI assistants","score":0.367000013589859}],"concepts":[{"id":"https://openalex.org/C71611378","wikidata":"https://www.wikidata.org/wiki/Q5165191","display_name":"Contextual design","level":3,"score":0.7717000246047974},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6811000108718872},{"id":"https://openalex.org/C2781265381","wikidata":"https://www.wikidata.org/wiki/Q5710255","display_name":"Helpfulness","level":2,"score":0.6240000128746033},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.5443999767303467},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.41519999504089355},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.4023999869823456},{"id":"https://openalex.org/C107457646","wikidata":"https://www.wikidata.org/wiki/Q207434","display_name":"Human\u2013computer interaction","level":1,"score":0.4000999927520752},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3986000120639801},{"id":"https://openalex.org/C2777541608","wikidata":"https://www.wikidata.org/wiki/Q5165194","display_name":"Contextual learning","level":2,"score":0.3806000053882599},{"id":"https://openalex.org/C2780910867","wikidata":"https://www.wikidata.org/wiki/Q1952416","display_name":"Multimodality","level":2,"score":0.3774999976158142},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.35920000076293945},{"id":"https://openalex.org/C135355979","wikidata":"https://www.wikidata.org/wiki/Q5165193","display_name":"Contextual inquiry","level":2,"score":0.33500000834465027},{"id":"https://openalex.org/C2780009758","wikidata":"https://www.wikidata.org/wiki/Q6804172","display_name":"Measure (data warehouse)","level":2,"score":0.3260999917984009},{"id":"https://openalex.org/C2777317252","wikidata":"https://www.wikidata.org/wiki/Q18393516","display_name":"Rare events","level":2,"score":0.32089999318122864},{"id":"https://openalex.org/C52085439","wikidata":"https://www.wikidata.org/wiki/Q5165173","display_name":"Context analysis","level":3,"score":0.3018999993801117},{"id":"https://openalex.org/C114938261","wikidata":"https://www.wikidata.org/wiki/Q1211272","display_name":"Signage","level":2,"score":0.28760001063346863},{"id":"https://openalex.org/C15744967","wikidata":"https://www.wikidata.org/wiki/Q9418","display_name":"Psychology","level":0,"score":0.28200000524520874},{"id":"https://openalex.org/C2778143727","wikidata":"https://www.wikidata.org/wiki/Q1820650","display_name":"Readability","level":2,"score":0.2759000062942505},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.27399998903274536}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48448/m2wa-5w92","doi":"https://doi.org/10.48448/m2wa-5w92","is_oa":true,"landing_page_url":"https://doi.org/10.48448/m2wa-5w92","pdf_url":null,"source":{"id":"https://openalex.org/S7407053148","display_name":"Underline Science Inc.","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":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Audiovisual"}],"best_oa_location":{"id":"doi:10.48448/m2wa-5w92","doi":"https://doi.org/10.48448/m2wa-5w92","is_oa":true,"landing_page_url":"https://doi.org/10.48448/m2wa-5w92","pdf_url":null,"source":{"id":"https://openalex.org/S7407053148","display_name":"Underline Science Inc.","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":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Audiovisual"},"sustainable_development_goals":[],"sustainable_development_goals_aurora":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W7138947587","https://openalex.org/W7168999506","https://openalex.org/W4403344692","https://openalex.org/W4416255190","https://openalex.org/W7119739586","https://openalex.org/W7213926419","https://openalex.org/W4414813469","https://openalex.org/W7222657424","https://openalex.org/W7117301373","https://openalex.org/W4407690497"],"abstract_inverted_index":{"Multimodal":[0,67],"large":[1],"language":[2],"models":[3,150],"(MLLMs)":[4],"are":[5,39],"increasingly":[6],"deployed":[7],"as":[8],"assistants":[9],"that":[10,80,118,169],"interact":[11],"through":[12],"text":[13],"and":[14,31,42,60,77,92,99,112,128,136,147,167],"images,":[15],"making":[16],"it":[17],"crucial":[18],"to":[19,152],"evaluate":[20,163],"contextual":[21,36,82,121,145,180],"safety":[22,37,83,146],"when":[23],"risk":[24,91],"depends":[25],"on":[26,125,130],"both":[27,58],"the":[28,32,53,65],"visual":[29],"scene":[30,55],"evolving":[33],"dialogue.":[34],"Existing":[35],"benchmarks":[38],"mostly":[40],"single-turn":[41],"often":[43],"miss":[44,154],"how":[45,52],"malicious":[46],"intent":[47,122],"can":[48,56],"emerge":[49],"gradually":[50],"or":[51,157],"same":[54],"support":[57],"benign":[59,131,159],"exploitative":[61],"goals.":[62],"We":[63],"introduce":[64],"Multi-Turn":[66],"Contextual":[68],"Safety":[69],"Benchmark":[70],"(MTMCS-Bench),":[71],"a":[72],"benchmark":[73],"of":[74],"realistic":[75],"images":[76],"multi-turn":[78,179],"conversations":[79],"evaluates":[81],"in":[84],"MLLMs":[85],"under":[86],"two":[87],"complementary":[88],"settings,":[89],"escalation-based":[90],"context-switch":[93],"risk.":[94],"MTMCS-Bench":[95],"offers":[96],"paired":[97],"safe":[98],"unsafe":[100,126],"dialogues":[101],"with":[102,116,149],"structured":[103],"evaluation.":[104],"It":[105],"contains":[106],"over":[107],"30":[108],"thousand":[109],"multimodal":[110],"(image+text)":[111],"unimodal":[113],"(text-only)":[114],"samples,":[115],"metrics":[117],"separately":[119],"measure":[120],"recognition,":[123],"safety-awareness":[124],"cases,":[127],"helpfulness":[129],"ones.":[132],"Across":[133],"eight":[134],"open-source":[135],"seven":[137],"proprietary":[138],"MLLMs,":[139],"we":[140,162],"observe":[141],"persistent":[142],"trade-offs":[143],"between":[144],"utility,":[148],"tending":[151],"either":[153],"gradual":[155],"risks":[156],"over-refuse":[158],"dialogues.":[160],"Finally,":[161],"five":[164],"current":[165],"guardrails":[166],"find":[168],"they":[170],"mitigate":[171],"some":[172],"failures":[173],"but":[174],"do":[175],"not":[176],"fully":[177],"resolve":[178],"risks.":[181]},"counts_by_year":[],"updated_date":"2026-10-28T10:33:24.094632","created_date":"2026-07-17T00:00:00"}