{"id":"https://openalex.org/W7166802255","doi":"https://doi.org/10.18653/v1/2026.acl-long.1168","title":"SafetyMem: Adaptive Jailbreak Defense via Dual-Component Safety Memory","display_name":"SafetyMem: Adaptive Jailbreak Defense via Dual-Component Safety Memory","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7166802255","doi":"https://doi.org/10.18653/v1/2026.acl-long.1168"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.acl-long.1168","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.1168","pdf_url":"https://aclanthology.org/2026.acl-long.1168.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.1168.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5139793910","display_name":"Hao Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hao Wang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5130288322","display_name":"Ziyi Ni","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ziyi Ni","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5122964996","display_name":"Huacan Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Huacan Wang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129649388","display_name":"Pin Lyu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Pin Lyu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5139734966","display_name":"Lei Sha","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lei Sha","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":0.79167997,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"25486","last_page":"25509"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.27000001072883606,"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":0.27000001072883606,"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/T11424","display_name":"Security and Verification in Computing","score":0.09459999948740005,"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/T13295","display_name":"Safety Systems Engineering in Autonomy","score":0.05530000105500221,"subfield":{"id":"https://openalex.org/subfields/2213","display_name":"Safety, Risk, Reliability and Quality"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/component","display_name":"Component (thermodynamics)","score":0.2596000134944916},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.2590000033378601},{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.23149999976158142},{"id":"https://openalex.org/keywords/task","display_name":"Task (project management)","score":0.22130000591278076},{"id":"https://openalex.org/keywords/action","display_name":"Action (physics)","score":0.21770000457763672}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5810999870300293},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.2597000002861023},{"id":"https://openalex.org/C168167062","wikidata":"https://www.wikidata.org/wiki/Q1117970","display_name":"Component (thermodynamics)","level":2,"score":0.2596000134944916},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.2590000033378601},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.24580000340938568},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.23149999976158142},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.22130000591278076},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.21930000185966492},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.21770000457763672},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.20919999480247498}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2026.acl-long.1168","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.1168","pdf_url":"https://aclanthology.org/2026.acl-long.1168.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.1168","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.1168","pdf_url":"https://aclanthology.org/2026.acl-long.1168.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":[{"id":"https://openalex.org/G4458316040","display_name":null,"funder_award_id":"KZ37117501","funder_id":"https://openalex.org/F4320333688","funder_display_name":"National Outstanding Youth Science Fund Project of National Natural Science Foundation of China"},{"id":"https://openalex.org/G5831522889","display_name":null,"funder_award_id":"62306024","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"},{"id":"https://openalex.org/F4320329860","display_name":"National Science and Technology Major Project","ror":null},{"id":"https://openalex.org/F4320333688","display_name":"National Outstanding Youth Science Fund Project of National Natural Science Foundation of China","ror":null},{"id":"https://openalex.org/F4320335787","display_name":"Fundamental Research Funds for the Central Universities","ror":null}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7166802255.pdf","grobid_xml":"https://content.openalex.org/works/W7166802255.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Current":[0],"defenses":[1],"for":[2],"Large":[3],"Language":[4],"Models":[5],"(LLMs)":[6],"often":[7],"suffer":[8],"from":[9,79],"a":[10,34,41,58],"\"memory":[11],"gap\":":[12],"parameter-modifying":[13],"methods":[14],"are":[15],"computationally":[16],"rigid,":[17],"while":[18,133],"inference-time":[19],"filters":[20],"cannot":[21],"retain":[22],"or":[23],"reuse":[24],"defense":[25],"knowledge":[26,60],"across":[27,142],"interactions.To":[28],"address":[29],"this,":[30],"we":[31,103],"propose":[32],"Safet-yMem,":[33],"novel":[35],"framework":[36],"that":[37,110,126],"secures":[38],"LLMs":[39],"through":[40],"dual-component":[42],"safety":[43],"memory":[44,107],"system.SafetyMem":[45],"consists":[46],"of":[47,62,75],"Semantic":[48],"Safety":[49,67],"Memory":[50,68],"(SSM),":[51],"which":[52,70],"consolidates":[53],"diverse":[54],"jailbreak":[55,123],"attempts":[56],"into":[57],"structured":[59],"base":[61],"attack":[63,130],"patterns,":[64],"and":[65,91,121,136],"Episodic":[66],"(ESM),":[69],"maintains":[71],"an":[72,105],"evolving":[73],"set":[74],"procedural":[76],"rules":[77],"refined":[78],"historical":[80],"detection":[81],"failures.Unlike":[82],"static":[83],"defenses,":[84],"Safe-tyMem":[85],"allows":[86],"the":[87],"model":[88],"to":[89,93,115],"\"remember\"":[90],"adapt":[92],"emerging":[94],"adversarial":[95,106],"strategies":[96],"without":[97],"parameter":[98],"retraining.To":[99],"further":[100],"enhance":[101],"robustness,":[102],"introduce":[104],"expansion":[108],"mechanism":[109],"proactively":[111],"generates":[112],"challenging":[113],"variants":[114],"solidify":[116],"these":[117],"memories.Experiments":[118],"on":[119],"standard":[120],"stealthy":[122],"benchmarks":[124],"show":[125],"SafetyMem":[127],"substantially":[128],"reduces":[129],"success":[131],"rates":[132],"preserving":[134],"efficiency":[135],"interpretability,":[137],"consistently":[138],"outperforming":[139],"state-of-the-art":[140],"baselines":[141],"multiple":[143],"LLMs.":[144]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2026-07-02T00:00:00"}