{"id":"https://openalex.org/W7166884242","doi":"https://doi.org/10.18653/v1/2026.findings-acl.1188","title":"Experience-Driven Multi-Agent Optimization for Black-Box Jailbreak Attacks on Large Language Models","display_name":"Experience-Driven Multi-Agent Optimization for Black-Box Jailbreak Attacks on Large Language Models","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7166884242","doi":"https://doi.org/10.18653/v1/2026.findings-acl.1188"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.findings-acl.1188","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1188","pdf_url":"https://aclanthology.org/2026.findings-acl.1188.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.1188.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5139762553","display_name":"Zhaoyang Han","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhaoyang Han","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139723546","display_name":"Yihe Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yihe Liu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139787780","display_name":"Kai Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kai Zhang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5139817613","display_name":"Ping Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ping Li","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":0.87253963,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"23729","last_page":"23747"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.20000000298023224,"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":0.20000000298023224,"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/T10764","display_name":"Privacy-Preserving Technologies in Data","score":0.08919999748468399,"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":0.06080000102519989,"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/identification","display_name":"Identification (biology)","score":0.3084999918937683},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.2605000138282776},{"id":"https://openalex.org/keywords/feature","display_name":"Feature (linguistics)","score":0.24070000648498535},{"id":"https://openalex.org/keywords/domain","display_name":"Domain (mathematical analysis)","score":0.23759999871253967},{"id":"https://openalex.org/keywords/matching","display_name":"Matching (statistics)","score":0.23520000278949738}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6110000014305115},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.38029998540878296},{"id":"https://openalex.org/C116834253","wikidata":"https://www.wikidata.org/wiki/Q2039217","display_name":"Identification (biology)","level":2,"score":0.3084999918937683},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.2605000138282776},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.24860000610351562},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.24070000648498535},{"id":"https://openalex.org/C36503486","wikidata":"https://www.wikidata.org/wiki/Q11235244","display_name":"Domain (mathematical analysis)","level":2,"score":0.23759999871253967},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.23520000278949738},{"id":"https://openalex.org/C2776036281","wikidata":"https://www.wikidata.org/wiki/Q48769818","display_name":"Constraint (computer-aided design)","level":2,"score":0.23309999704360962},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.23229999840259552}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2026.findings-acl.1188","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1188","pdf_url":"https://aclanthology.org/2026.findings-acl.1188.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.1188","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1188","pdf_url":"https://aclanthology.org/2026.findings-acl.1188.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":[{"score":0.6246005296707153,"display_name":"Peace, Justice and strong institutions","id":"https://metadata.un.org/sdg/16"}],"awards":[{"id":"https://openalex.org/G4186086668","display_name":null,"funder_award_id":"No. 62276099","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G6304150607","display_name":null,"funder_award_id":"62276099","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/W7166884242.pdf","grobid_xml":"https://content.openalex.org/works/W7166884242.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"The":[0],"rapid":[1],"discovery":[2],"of":[3,11],"jailbreak":[4,20,130,146,184],"prompts":[5,116],"has":[6],"revealed":[7],"the":[8,79,160],"alarming":[9],"fragility":[10],"safety":[12,29],"alignment":[13],"in":[14,26,154],"frontier":[15],"large":[16],"language":[17],"models":[18],"(LLMs).While":[19],"techniques":[21],"play":[22],"a":[23,103,109],"critical":[24],"role":[25],"red-teaming":[27],"and":[28,61,64,100,117,123,128,179],"evaluation,":[30],"existing":[31,144],"methods":[32],"exhibit":[33],"three":[34,95],"key":[35],"limitations:":[36],"(i)":[37],"poor":[38,62],"transferability":[39],"across":[40,121,168],"model":[41,81],"families,":[42],"requiring":[43],"model-specific":[44],"manual":[45],"tuning;":[46],"(ii)":[47],"heavy":[48],"reliance":[49],"on":[50,136],"large-scale":[51],"prompt":[52],"enumeration":[53],"or":[54,71],"exhaustive":[55],"search,":[56],"causing":[57],"prohibitive":[58],"query":[59,162],"costs":[60],"scalability;":[63],"(iii)":[65],"high":[66],"sensitivity":[67],"to":[68,75,150,166],"input":[69],"preprocessing":[70],"refusal-oriented":[72],"fine-tuning,":[73],"leading":[74],"attack":[76,155],"failures":[77],"once":[78],"underlying":[80],"is":[82],"updated.To":[83],"address":[84],"these,":[85],"we":[86],"propose":[87],"Experience-driven":[88],"Multi-agent":[89],"Jailbreak":[90],"Optimization":[91],"(EMJO)":[92],"which":[93],"couples":[94],"collaborating":[96],"agents":[97],"(Attacker,":[98],"Analyzer,":[99],"Judge)":[101],"into":[102],"closedloop":[104],"\"probe-evaluate-revise\"":[105],"process,":[106],"together":[107],"with":[108],"dynamic":[110],"experience":[111],"bank":[112],"accumulating":[113],"high-quality":[114],"successful":[115],"reusable":[118],"strategy":[119],"patterns":[120],"iterations":[122],"tasks.This":[124],"design":[125],"enables":[126],"query-efficient":[127],"transferable":[129],"optimization":[131],"under":[132],"black-box":[133,145],"access.Extensive":[134],"experiments":[135],"diverse":[137],"LLMs":[138],"demonstrate":[139],"that":[140,174],"EMJO":[141,175],"consistently":[142],"outperforms":[143],"baselines,":[147],"achieving":[148],"up":[149,165],"11%":[151],"absolute":[152],"improvement":[153],"success":[156],"rate":[157],"while":[158],"reducing":[159],"average":[161],"cost":[163],"by":[164],"7.9\u00d7":[167],"two":[169],"benchmark":[170],"datasets.These":[171],"results":[172],"indicate":[173],"offers":[176],"an":[177],"effective":[178],"scalable":[180],"paradigm":[181],"for":[182],"systematic":[183],"discovery.":[185]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2026-07-02T00:00:00"}