{"id":"https://openalex.org/W4412944491","doi":"https://doi.org/10.18653/v1/2025.findings-acl.1311","title":"Red Queen: Exposing Latent Multi-Turn Risks in Large Language Models","display_name":"Red Queen: Exposing Latent Multi-Turn Risks in Large Language Models","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4412944491","doi":"https://doi.org/10.18653/v1/2025.findings-acl.1311"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.findings-acl.1311","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.1311","pdf_url":"https://aclanthology.org/2025.findings-acl.1311.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 2025","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.findings-acl.1311.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5101505105","display_name":"Yifan Jiang","orcid":"https://orcid.org/0009-0007-5334-9627"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yifan Jiang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5036907475","display_name":"Kriti Aggarwal","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kriti Aggarwal","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5114420685","display_name":"Tanmay Laud","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tanmay Laud","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5114420686","display_name":"Kashif Munir","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kashif Munir","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5073443117","display_name":"Jay Pujara","orcid":"https://orcid.org/0000-0001-6921-1744"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jay Pujara","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5118646350","display_name":"Subhabrata Mukherjee","orcid":"https://orcid.org/0009-0004-6684-4158"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Subhabrata Mukherjee","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":null,"has_fulltext":true,"cited_by_count":1,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"25554","last_page":"25591"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.9810000061988831,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.9810000061988831,"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/T10028","display_name":"Topic Modeling","score":0.9775000214576721,"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/computer-science","display_name":"Computer science","score":0.5151016116142273},{"id":"https://openalex.org/keywords/queen","display_name":"Queen (butterfly)","score":0.4550044536590576},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.4275047779083252},{"id":"https://openalex.org/keywords/natural-language-processing","display_name":"Natural language processing","score":0.27450329065322876}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5151016116142273},{"id":"https://openalex.org/C2781321516","wikidata":"https://www.wikidata.org/wiki/Q310836","display_name":"Queen (butterfly)","level":3,"score":0.4550044536590576},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.4275047779083252},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.27450329065322876},{"id":"https://openalex.org/C59822182","wikidata":"https://www.wikidata.org/wiki/Q441","display_name":"Botany","level":1,"score":0.0},{"id":"https://openalex.org/C86803240","wikidata":"https://www.wikidata.org/wiki/Q420","display_name":"Biology","level":0,"score":0.0},{"id":"https://openalex.org/C2780653484","wikidata":"https://www.wikidata.org/wiki/Q22651","display_name":"Hymenoptera","level":2,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.findings-acl.1311","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.1311","pdf_url":"https://aclanthology.org/2025.findings-acl.1311.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 2025","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.findings-acl.1311","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.1311","pdf_url":"https://aclanthology.org/2025.findings-acl.1311.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 2025","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4412944491.pdf","grobid_xml":"https://content.openalex.org/works/W4412944491.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W649410576","https://openalex.org/W4229490663","https://openalex.org/W4250559856","https://openalex.org/W4256613351","https://openalex.org/W635590649","https://openalex.org/W4235414885","https://openalex.org/W2809069473"],"abstract_inverted_index":{"Content":[0],"Warning:":[1],"This":[2],"paper":[3],"contains":[4],"examples":[5],"of":[6,13,71,111,131,169,196],"harmful":[7,36,125],"language":[8,15],"and":[9,24,62,123,151,199,225],"plans.The":[10],"rapid":[11],"advancement":[12],"large":[14],"models":[16,136,184],"(LLMs)":[17],"has":[18,27],"unlocked":[19],"diverse":[20],"opportunities":[21],"across":[22,120],"domains":[23],"applications":[25],"but":[26],"also":[28],"raised":[29],"concerns":[30],"about":[31],"their":[32,66],"tendency":[33],"to":[34,52,79,139,155,178,188,193,214],"generate":[35,114],"responses":[37],"under":[38,108],"jailbreak":[39,43,94,157],"attacks.However,":[40],"most":[41],"existing":[42],"strategies":[44],"are":[45,137,227],"singleturn":[46],"with":[47,172],"explicit":[48],"malicious":[49,106],"intent,":[50],"failing":[51],"reflect":[53],"the":[54,105,109,159,194],"real-world":[55],"scenario":[56],"where":[57],"interactions":[58],"can":[59,64],"be":[60],"multi-turn":[61,102,116,197],"users":[63],"conceal":[65],"intents.Recent":[67],"studies":[68],"on":[69,87,149,153,166,221],"Theory":[70],"Mind":[72],"(ToM)":[73],"reveal":[74],"that":[75,182],"LLMs":[76],"often":[77],"struggle":[78],"infer":[80],"users'":[81],"latent":[82],"intent":[83,107],"in":[84],"such":[85],"scenarios.Building":[86],"these":[88],"limitations,":[89],"we":[90,204],"propose":[91,205],"a":[92,101,209],"novel":[93],"attack,":[95,190],"RED":[96,140,160,206],"QUEEN":[97,141,161,207],"ATTACK,":[98,142],"which":[99],"constructs":[100],"scenario,":[103],"concealing":[104],"guise":[110],"preventing":[112],"harm.We":[113],"56k":[115],"concealment":[117,200],"data":[118],"points":[119],"40":[121],"scenarios":[122],"14":[124],"categories,":[126],"evaluating":[127],"four":[128],"LLM":[129],"families":[130],"different":[132],"sizes.Results":[133],"show":[134],"all":[135],"vulnerable":[138],"reaching":[143],"87.6%":[144],"attack":[145],"success":[146],"rate":[147],"(ASR)":[148],"GPT-4o":[150],"77.1%":[152],"Llama3-70B.Compared":[154],"prior":[156],"attacks,":[158],"ATTACK":[162],"achieves":[163],"superior":[164],"performance":[165,220],"nine":[167],"out":[168],"ten":[170],"models,":[171],"ASR":[173,213],"improvements":[174],"ranging":[175],"from":[176],"2%":[177],"64%.Further":[179],"analysis":[180],"reveals":[181],"larger":[183],"exhibit":[185],"greater":[186],"vulnerability":[187],"our":[189],"primarily":[191],"due":[192],"combination":[195],"structures":[198],"strategies.To":[201],"enhance":[202],"safety,":[203],"GUARD,":[208],"mitigation":[210],"strategy":[211],"reducing":[212],"below":[215],"1%":[216],"while":[217],"maintaining":[218],"model":[219],"standard":[222],"benchmarks.Full":[223],"implementation":[224],"dataset":[226],"publicly":[228],"accessible":[229],"at":[230],"https://github.com/kriti-hippo/":[231],"red_queen.":[232]},"counts_by_year":[{"year":2026,"cited_by_count":1}],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}