{"id":"https://openalex.org/W7131278703","doi":"https://doi.org/10.1109/acsac67867.2025.00095","title":"Siren: A Learning-Based Multi-Turn Attack Framework for Simulating Real-World Human Jailbreak Behaviors","display_name":"Siren: A Learning-Based Multi-Turn Attack Framework for Simulating Real-World Human Jailbreak Behaviors","publication_year":2025,"publication_date":"2025-12-08","ids":{"openalex":"https://openalex.org/W7131278703","doi":"https://doi.org/10.1109/acsac67867.2025.00095"},"language":null,"primary_location":{"id":"doi:10.1109/acsac67867.2025.00095","is_oa":false,"landing_page_url":"https://doi.org/10.1109/acsac67867.2025.00095","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2025 IEEE Annual Computer Security Applications Conference (ACSAC)","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":false,"oa_status":"closed","oa_url":null,"any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5102930048","display_name":"Yi Zhao","orcid":"https://orcid.org/0000-0002-2108-0463"},"institutions":[{"id":"https://openalex.org/I14243506","display_name":"Hong Kong Polytechnic University","ror":"https://ror.org/0030zas98","country_code":"HK","type":"education","lineage":["https://openalex.org/I14243506"]}],"countries":["HK"],"is_corresponding":true,"raw_author_name":"Yi Zhao","raw_affiliation_strings":["The Hong Kong Polytechnic University,Department of Computing,Hong Kong SAR,China"],"affiliations":[{"raw_affiliation_string":"The Hong Kong Polytechnic University,Department of Computing,Hong Kong SAR,China","institution_ids":["https://openalex.org/I14243506"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5116039137","display_name":"Youzhi Zhang","orcid":null},"institutions":[{"id":"https://openalex.org/I4210121405","display_name":"Centre for Artificial Intelligence and Robotics","ror":"https://ror.org/01xnbq218","country_code":"IN","type":"facility","lineage":["https://openalex.org/I1340206300","https://openalex.org/I4210121405","https://openalex.org/I4210150591"]}],"countries":["IN"],"is_corresponding":false,"raw_author_name":"Youzhi Zhang","raw_affiliation_strings":["Hong Kong Institute of Science & Innovation Chinese Academy of Sciences,Centre for Artificial Intelligence and Robotics,Hong Kong SAR,China"],"affiliations":[{"raw_affiliation_string":"Hong Kong Institute of Science & Innovation Chinese Academy of Sciences,Centre for Artificial Intelligence and Robotics,Hong Kong SAR,China","institution_ids":["https://openalex.org/I4210121405"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":2,"corresponding_author_ids":["https://openalex.org/A5102930048"],"corresponding_institution_ids":["https://openalex.org/I14243506"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.81510205,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"1209","last_page":"1220"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10565","display_name":"Psychopathy, Forensic Psychiatry, Sexual Offending","score":0.1298000067472458,"subfield":{"id":"https://openalex.org/subfields/3203","display_name":"Clinical Psychology"},"field":{"id":"https://openalex.org/fields/32","display_name":"Psychology"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},"topics":[{"id":"https://openalex.org/T10565","display_name":"Psychopathy, Forensic Psychiatry, Sexual Offending","score":0.1298000067472458,"subfield":{"id":"https://openalex.org/subfields/3203","display_name":"Clinical Psychology"},"field":{"id":"https://openalex.org/fields/32","display_name":"Psychology"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},{"id":"https://openalex.org/T13582","display_name":"Stalking, Cyberstalking, and Harassment","score":0.10689999908208847,"subfield":{"id":"https://openalex.org/subfields/3312","display_name":"Sociology and Political Science"},"field":{"id":"https://openalex.org/fields/33","display_name":"Social Sciences"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},{"id":"https://openalex.org/T10574","display_name":"Crime Patterns and Interventions","score":0.09780000150203705,"subfield":{"id":"https://openalex.org/subfields/3312","display_name":"Sociology and Political Science"},"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/siren","display_name":"Siren (mythology)","score":0.946399986743927},{"id":"https://openalex.org/keywords/set","display_name":"Set (abstract data type)","score":0.617900013923645},{"id":"https://openalex.org/keywords/focus","display_name":"Focus (optics)","score":0.5674999952316284},{"id":"https://openalex.org/keywords/baseline","display_name":"Baseline (sea)","score":0.4474000036716461},{"id":"https://openalex.org/keywords/code","display_name":"Code (set theory)","score":0.42480000853538513},{"id":"https://openalex.org/keywords/exploit","display_name":"Exploit","score":0.42010000348091125}],"concepts":[{"id":"https://openalex.org/C160844653","wikidata":"https://www.wikidata.org/wiki/Q150986","display_name":"Siren (mythology)","level":2,"score":0.946399986743927},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6786999702453613},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.617900013923645},{"id":"https://openalex.org/C192209626","wikidata":"https://www.wikidata.org/wiki/Q190909","display_name":"Focus (optics)","level":2,"score":0.5674999952316284},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.4627000093460083},{"id":"https://openalex.org/C12725497","wikidata":"https://www.wikidata.org/wiki/Q810247","display_name":"Baseline (sea)","level":2,"score":0.4474000036716461},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.42480000853538513},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.42010000348091125},{"id":"https://openalex.org/C124681953","wikidata":"https://www.wikidata.org/wiki/Q339062","display_name":"Decomposition","level":2,"score":0.3776000142097473},{"id":"https://openalex.org/C2781249084","wikidata":"https://www.wikidata.org/wiki/Q908656","display_name":"Preference","level":2,"score":0.337799996137619},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3278000056743622},{"id":"https://openalex.org/C2779639559","wikidata":"https://www.wikidata.org/wiki/Q7661178","display_name":"Symbolic execution","level":3,"score":0.2705000042915344},{"id":"https://openalex.org/C2780589192","wikidata":"https://www.wikidata.org/wiki/Q7285140","display_name":"Raising (metalworking)","level":2,"score":0.2619999945163727},{"id":"https://openalex.org/C95713431","wikidata":"https://www.wikidata.org/wiki/Q631425","display_name":"Vulnerability (computing)","level":2,"score":0.2549999952316284},{"id":"https://openalex.org/C107457646","wikidata":"https://www.wikidata.org/wiki/Q207434","display_name":"Human\u2013computer interaction","level":1,"score":0.2500999867916107}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/acsac67867.2025.00095","is_oa":false,"landing_page_url":"https://doi.org/10.1109/acsac67867.2025.00095","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2025 IEEE Annual Computer Security Applications Conference (ACSAC)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/16","display_name":"Peace, Justice and strong institutions","score":0.5102974772453308}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":15,"referenced_works":["https://openalex.org/W4389523689","https://openalex.org/W4393157467","https://openalex.org/W4401042259","https://openalex.org/W4401042853","https://openalex.org/W4401043746","https://openalex.org/W4401863316","https://openalex.org/W4402669751","https://openalex.org/W4402671039","https://openalex.org/W4402671718","https://openalex.org/W4402672027","https://openalex.org/W4404782248","https://openalex.org/W4404971291","https://openalex.org/W4405181600","https://openalex.org/W4410609100","https://openalex.org/W7127179963"],"related_works":[],"abstract_inverted_index":{"Large":[0],"language":[1],"models":[2],"(LLMs)":[3],"are":[4,173],"widely":[5],"used":[6,38],"in":[7],"real-world":[8,40,73],"applications,":[9],"raising":[10],"concerns":[11],"about":[12],"their":[13],"safety":[14],"and":[15,98,103,109,135,168],"trustworthiness.":[16],"While":[17],"red-teaming":[18],"with":[19,94,124,137,147,177],"jailbreak":[20,75,192],"prompts":[21],"exposes":[22],"the":[23,35,57,107,127,132,162,184],"vulnerabilities":[24],"of":[25,79,122,186],"LLMs,":[26],"current":[27],"efforts":[28],"focus":[29],"primarily":[30],"on":[31,46],"single-turn":[32,143],"attacks,":[33],"overlooking":[34],"multi-turn":[36,43,67,156,191],"strategies":[37,59,171],"by":[39],"adversaries.":[41],"Existing":[42],"methods":[44],"rely":[45],"static":[47],"patterns":[48],"or":[49],"predefined":[50],"logical":[51],"chains,":[52],"failing":[53],"to":[54,71,154],"account":[55],"for":[56],"dynamic":[58],"during":[60],"attacks.":[61],"We":[62,180],"propose":[63],"Siren,":[64],"a":[65,148,155],"learning-based":[66],"attack":[68,118,178],"framework":[69],"designed":[70],"simulate":[72],"human":[74],"behaviors.":[76],"Siren":[77,115,146,182],"consists":[78],"three":[80],"stages:":[81],"(1)":[82],"MiniMax-driven":[83],"training":[84],"set":[85],"construction":[86],"utilizing":[87],"Turn-Level":[88],"LLM":[89],"feedback,":[90],"(2)":[91],"post-training":[92],"attackers":[93],"supervised":[95],"fine-tuning":[96],"(SFT)":[97],"direct":[99],"preference":[100],"optimization":[101],"(DPO),":[102],"(3)":[104],"interactions":[105],"between":[106],"attacking":[108],"target":[110,133],"LLMs.":[111],"Experiments":[112],"demonstrate":[113],"that":[114,158,172],"achieves":[116,151],"an":[117],"success":[119],"rate":[120],"(ASR)":[121],"90%":[123],"LLaMA-3-8B":[125],"as":[126,131,161],"attacker":[128],"against":[129,139,189],"Gemini-1.5-Pro":[130],"model,":[134],"70%":[136],"Mistral-7B":[138],"GPT-4o,":[140],"significantly":[141],"outperforming":[142],"baselines.":[144],"Moreover,":[145],"7B-scale":[149],"model":[150],"performance":[152],"comparable":[153],"baseline":[157],"leverages":[159],"GPT-4o":[160],"attacker,":[163],"while":[164],"requiring":[165],"fewer":[166],"turns":[167],"employing":[169],"decomposition":[170],"better":[174],"semantically":[175],"aligned":[176],"goals.":[179],"hope":[181],"inspires":[183],"development":[185],"stronger":[187],"defenses":[188],"advanced":[190],"attacks":[193],"under":[194],"realistic":[195],"scenarios.":[196],"Code":[197],"is":[198],"available":[199],"at":[200],"https://github.com/YiyiyiZhao/siren.":[201],"Warning:":[202],"This":[203],"paper":[204],"contains":[205],"potentially":[206],"harmful":[207],"text.":[208]},"counts_by_year":[],"updated_date":"2026-02-25T21:11:00.739837","created_date":"2026-02-25T00:00:00"}