{"id":"https://openalex.org/W7166824033","doi":"https://doi.org/10.18653/v1/2026.acl-long.1895","title":"Retrieval-Augmented Defense: Adaptive and Controllable Jailbreak Prevention for Large Language Models","display_name":"Retrieval-Augmented Defense: Adaptive and Controllable Jailbreak Prevention for Large Language Models","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7166824033","doi":"https://doi.org/10.18653/v1/2026.acl-long.1895"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.acl-long.1895","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.1895","pdf_url":"https://aclanthology.org/2026.acl-long.1895.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":"Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2026.acl-long.1895.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5139794253","display_name":"Guangyu Yang","orcid":null},"institutions":[{"id":"https://openalex.org/I241749","display_name":"University of Cambridge","ror":"https://ror.org/013meh722","country_code":"GB","type":"education","lineage":["https://openalex.org/I241749"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Guangyu Yang","raw_affiliation_strings":["Department of Engineering University of Cambridge"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Engineering University of Cambridge","institution_ids":["https://openalex.org/I241749"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139764876","display_name":"Jinghong Chen","orcid":null},"institutions":[{"id":"https://openalex.org/I241749","display_name":"University of Cambridge","ror":"https://ror.org/013meh722","country_code":"GB","type":"education","lineage":["https://openalex.org/I241749"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Jinghong Chen","raw_affiliation_strings":["Department of Engineering University of Cambridge"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Engineering University of Cambridge","institution_ids":["https://openalex.org/I241749"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5071561232","display_name":"Jingbiao Mei","orcid":null},"institutions":[{"id":"https://openalex.org/I241749","display_name":"University of Cambridge","ror":"https://ror.org/013meh722","country_code":"GB","type":"education","lineage":["https://openalex.org/I241749"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Jingbiao Mei","raw_affiliation_strings":["Department of Engineering University of Cambridge"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Engineering University of Cambridge","institution_ids":["https://openalex.org/I241749"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5051401703","display_name":"Weizhe Lin","orcid":"https://orcid.org/0000-0002-0754-4524"},"institutions":[{"id":"https://openalex.org/I241749","display_name":"University of Cambridge","ror":"https://ror.org/013meh722","country_code":"GB","type":"education","lineage":["https://openalex.org/I241749"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Weizhe Lin","raw_affiliation_strings":["Department of Engineering University of Cambridge"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Engineering University of Cambridge","institution_ids":["https://openalex.org/I241749"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5139743633","display_name":"Bill Byrne","orcid":null},"institutions":[{"id":"https://openalex.org/I241749","display_name":"University of Cambridge","ror":"https://ror.org/013meh722","country_code":"GB","type":"education","lineage":["https://openalex.org/I241749"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Bill Byrne","raw_affiliation_strings":["Department of Engineering University of Cambridge"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Engineering University of Cambridge","institution_ids":["https://openalex.org/I241749"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I241749"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.8133844,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"40849","last_page":"40868"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.20550000667572021,"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":0.20550000667572021,"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":0.17020000517368317,"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":0.05380000174045563,"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/control","display_name":"Control (management)","score":0.30709999799728394},{"id":"https://openalex.org/keywords/adaptive-control","display_name":"Adaptive control","score":0.2874000072479248},{"id":"https://openalex.org/keywords/class","display_name":"Class (philosophy)","score":0.27230000495910645},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.2632000148296356},{"id":"https://openalex.org/keywords/identification","display_name":"Identification (biology)","score":0.26269999146461487}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6141999959945679},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4131999909877777},{"id":"https://openalex.org/C2775924081","wikidata":"https://www.wikidata.org/wiki/Q55608371","display_name":"Control (management)","level":2,"score":0.30709999799728394},{"id":"https://openalex.org/C107464732","wikidata":"https://www.wikidata.org/wiki/Q235781","display_name":"Adaptive control","level":3,"score":0.2874000072479248},{"id":"https://openalex.org/C2777212361","wikidata":"https://www.wikidata.org/wiki/Q5127848","display_name":"Class (philosophy)","level":2,"score":0.27230000495910645},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.2632000148296356},{"id":"https://openalex.org/C116834253","wikidata":"https://www.wikidata.org/wiki/Q2039217","display_name":"Identification (biology)","level":2,"score":0.26269999146461487},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.25999999046325684},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.2513999938964844},{"id":"https://openalex.org/C17500928","wikidata":"https://www.wikidata.org/wiki/Q959968","display_name":"Control system","level":2,"score":0.2475000023841858}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2026.acl-long.1895","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.1895","pdf_url":"https://aclanthology.org/2026.acl-long.1895.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":"Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.acl-long.1895","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.1895","pdf_url":"https://aclanthology.org/2026.acl-long.1895.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":"Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[{"id":"https://openalex.org/F4320320279","display_name":"University of Warwick","ror":"https://ror.org/01a77tt86"},{"id":"https://openalex.org/F4320321848","display_name":"Cambridge Trust","ror":"https://ror.org/05sprqy15"},{"id":"https://openalex.org/F4320334639","display_name":"Christ's College, University of Cambridge","ror":null}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7166824033.pdf","grobid_xml":"https://content.openalex.org/works/W7166824033.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Large":[0],"Language":[1],"Models":[2],"(LLMs)":[3],"remain":[4],"vulnerable":[5],"to":[6,11,33,79,90,106],"jailbreak":[7,63,87,100,122,166],"attacks,":[8],"which":[9,76],"attempt":[10],"elicit":[12],"harmful":[13,165],"responses":[14],"from":[15],"LLMs.The":[16],"evolving":[17],"nature":[18],"and":[19,41,49,86,102,109,127,142],"diversity":[20],"of":[21,44,69,120,154],"these":[22,52],"attacks":[23,123],"pose":[24],"many":[25],"challenges":[26],"for":[27,62,97,134,168],"defense":[28],"systems,":[29],"including":[30],"(1)":[31],"adaptation":[32],"counter":[34],"emerging":[35],"attack":[36,71,91],"strategies":[37,101],"without":[38],"costly":[39],"retraining,":[40],"(2)":[42],"control":[43],"the":[45,81,92,118],"trade-off":[46,150],"between":[47],"safety":[48,108],"utility.To":[50],"address":[51],"challenges,":[53],"we":[54],"propose":[55,137],"Retrieval-Augmented":[56,74],"Defense":[57],"(RAD),":[58],"a":[59,67,104,138,147,152,158],"novel":[60,139],"framework":[61],"detection":[64],"that":[65,114,144,171],"incorporates":[66],"database":[68],"known":[70],"examples":[72],"into":[73],"Generation,":[75],"is":[77],"used":[78,89],"infer":[80],"underlying,":[82],"malicious":[83],"user":[84],"query":[85],"strategy":[88],"system.RAD":[93],"enables":[94],"training-free":[95],"updates":[96],"newly":[98],"discovered":[99],"provides":[103],"mechanism":[105],"balance":[107],"utility.Experiments":[110],"on":[111],"StrongREJECT":[112],"show":[113,143],"RAD":[115,145],"substantially":[116],"reduces":[117],"effectiveness":[119],"strong":[121],"such":[124],"as":[125],"PAP":[126],"PAIR":[128],"while":[129],"maintaining":[130],"low":[131],"rejection":[132],"rates":[133],"benign":[135],"queries.We":[136],"evaluation":[140],"scheme":[141],"achieves":[146],"robust":[148],"safety-utility":[149],"across":[151],"range":[153],"operating":[155],"points":[156],"in":[157],"controllable":[159],"manner.":[160],"1":[161],"This":[162],"paper":[163],"contains":[164],"contents":[167],"demonstration":[169],"purposes":[170],"can":[172],"be":[173],"offensive.":[174]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2026-07-02T00:00:00"}
