{"id":"https://openalex.org/W4412888672","doi":"https://doi.org/10.18653/v1/2025.findings-acl.325","title":"Why Not Act on What You Know? Unleashing Safety Potential of LLMs via Self-Aware Guard Enhancement","display_name":"Why Not Act on What You Know? Unleashing Safety Potential of LLMs via Self-Aware Guard Enhancement","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4412888672","doi":"https://doi.org/10.18653/v1/2025.findings-acl.325"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.findings-acl.325","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.325","pdf_url":"https://aclanthology.org/2025.findings-acl.325.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.325.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5102579940","display_name":"Peng Ding","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Peng Ding","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Jun Kuang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jun Kuang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"ZongYu Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"ZongYu Wang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5108060642","display_name":"Xuezhi Cao","orcid":"https://orcid.org/0000-0002-7044-1341"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xuezhi Cao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5068562468","display_name":"Xunliang Cai","orcid":"https://orcid.org/0000-0003-0685-9556"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xunliang Cai","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100427080","display_name":"Jiajun Chen","orcid":"https://orcid.org/0000-0003-2534-6633"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jiajun Chen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5102865824","display_name":"Shujian Huang","orcid":"https://orcid.org/0000-0003-4869-0832"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shujian Huang","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":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"6279","last_page":"6299"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10917","display_name":"Smart Grid Security and Resilience","score":0.8241999745368958,"subfield":{"id":"https://openalex.org/subfields/2207","display_name":"Control and Systems Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T10917","display_name":"Smart Grid Security and Resilience","score":0.8241999745368958,"subfield":{"id":"https://openalex.org/subfields/2207","display_name":"Control and Systems Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10270","display_name":"Blockchain Technology Applications and Security","score":0.8125,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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/T11891","display_name":"Big Data and Business Intelligence","score":0.7580000162124634,"subfield":{"id":"https://openalex.org/subfields/1404","display_name":"Management Information Systems"},"field":{"id":"https://openalex.org/fields/14","display_name":"Business, Management and Accounting"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/guard","display_name":"Guard (computer science)","score":0.7796292304992676},{"id":"https://openalex.org/keywords/business","display_name":"Business","score":0.49321597814559937},{"id":"https://openalex.org/keywords/computer-security","display_name":"Computer security","score":0.4931339621543884},{"id":"https://openalex.org/keywords/coast-guard","display_name":"Coast guard","score":0.4757639765739441},{"id":"https://openalex.org/keywords/risk-analysis","display_name":"Risk analysis (engineering)","score":0.4316936731338501},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.3823182284832001},{"id":"https://openalex.org/keywords/engineering","display_name":"Engineering","score":0.1767069399356842}],"concepts":[{"id":"https://openalex.org/C141141315","wikidata":"https://www.wikidata.org/wiki/Q2379942","display_name":"Guard (computer science)","level":2,"score":0.7796292304992676},{"id":"https://openalex.org/C144133560","wikidata":"https://www.wikidata.org/wiki/Q4830453","display_name":"Business","level":0,"score":0.49321597814559937},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.4931339621543884},{"id":"https://openalex.org/C3017894346","wikidata":"https://www.wikidata.org/wiki/Q1353040","display_name":"Coast guard","level":2,"score":0.4757639765739441},{"id":"https://openalex.org/C112930515","wikidata":"https://www.wikidata.org/wiki/Q4389547","display_name":"Risk analysis (engineering)","level":1,"score":0.4316936731338501},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.3823182284832001},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.1767069399356842},{"id":"https://openalex.org/C199104240","wikidata":"https://www.wikidata.org/wiki/Q118291","display_name":"Marine engineering","level":1,"score":0.0},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.findings-acl.325","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.325","pdf_url":"https://aclanthology.org/2025.findings-acl.325.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.325","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.325","pdf_url":"https://aclanthology.org/2025.findings-acl.325.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":[{"id":"https://openalex.org/F4320306076","display_name":"National Science Foundation","ror":"https://ror.org/021nxhr62"},{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"},{"id":"https://openalex.org/F4320335787","display_name":"Fundamental Research Funds for the Central Universities","ror":null}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4412888672.pdf","grobid_xml":"https://content.openalex.org/works/W4412888672.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W3018021567","https://openalex.org/W2797800246","https://openalex.org/W624352650","https://openalex.org/W2909860811","https://openalex.org/W607404756","https://openalex.org/W624921809","https://openalex.org/W3023570606","https://openalex.org/W4243136610","https://openalex.org/W4399685216"],"abstract_inverted_index":{"Large":[0],"Language":[1],"Models":[2],"(LLMs)":[3],"have":[4],"shown":[5],"impressive":[6],"capabilities":[7],"across":[8,104],"various":[9,105],"tasks":[10],"but":[11],"remain":[12],"vulnerable":[13],"to":[14,59,157],"meticulously":[15],"crafted":[16],"jailbreak":[17,33,91,127],"attacks.In":[18],"this":[19,46,151],"paper,":[20],"we":[21,48],"identify":[22],"a":[23,54,78,83],"critical":[24],"safety":[25,63,70,95,163],"gap:":[26],"while":[27,129],"LLMs":[28,109,160],"are":[29,171],"adept":[30],"at":[31,174],"detecting":[32],"prompts,":[34],"they":[35],"often":[36],"produce":[37],"unsafe":[38],"responses":[39],"when":[40],"directly":[41],"processing":[42],"these":[43],"inputs.Inspired":[44],"by":[45],"insight,":[47],"propose":[49],"SAGE":[50],"(Self-Aware":[51],"Guard":[52],"Enhancement),":[53],"training-free":[55],"defense":[56,119],"strategy":[57],"designed":[58],"align":[60],"LLMs'":[61],"strong":[62],"discrimination":[64,96],"performance":[65],"with":[66,161],"their":[67],"relatively":[68],"weaker":[69],"generation":[71,166],"ability.SAGE":[72],"consists":[73],"of":[74,110,150],"two":[75],"core":[76],"components:":[77],"Discriminative":[79,84],"Analysis":[80],"Module":[81],"and":[82,102,107,113,125,143,165,169],"Response":[85],"Module,":[86],"enhancing":[87],"resilience":[88],"against":[89,122],"sophisticated":[90],"attempts":[92],"through":[93,140],"flexible":[94],"instructions.Extensive":[97],"experiments":[98],"demonstrate":[99],"SAGE's":[100],"effectiveness":[101],"robustness":[103],"open-source":[106],"closed-source":[108],"different":[111],"sizes":[112],"architectures,":[114],"achieving":[115],"an":[116],"average":[117],"99%":[118],"success":[120],"rate":[121],"numerous":[123],"complex":[124],"covert":[126],"methods":[128],"maintaining":[130],"helpfulness":[131],"on":[132],"general":[133],"benchmarks.We":[134],"further":[135],"conduct":[136],"mechanistic":[137],"interpretability":[138],"analysis":[139],"hidden":[141],"states":[142],"attention":[144],"distributions,":[145],"revealing":[146],"the":[147],"underlying":[148],"mechanisms":[149],"detection-generation":[152],"discrepancy.Our":[153],"work":[154],"thus":[155],"contributes":[156],"developing":[158],"future":[159],"coherent":[162],"awareness":[164],"behavior.Our":[167],"code":[168],"datasets":[170],"publicly":[172],"available":[173],"https:":[175],"//github.com/NJUNLP/SAGE.":[176]},"counts_by_year":[],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}