{"id":"https://openalex.org/W7220686358","doi":"https://doi.org/10.18653/2026.emnlp-main.1549","title":"Reactivating Alignment: Defending LLMs from Jailbreaks via Intention-Aware Input\u2013Output Matching","display_name":"Reactivating Alignment: Defending LLMs from Jailbreaks via Intention-Aware Input\u2013Output Matching","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7220686358","doi":"https://doi.org/10.18653/2026.emnlp-main.1549"},"language":"en","primary_location":{"id":"doi:10.18653/2026.emnlp-main.1549","doi":"https://doi.org/10.18653/2026.emnlp-main.1549","is_oa":true,"landing_page_url":"https://doi.org/10.18653/2026.emnlp-main.1549","pdf_url":"https://aclanthology.org/2026.emnlp-main.1549.pdf","source":{"id":"https://openalex.org/S4363608991","display_name":"Conference on Empirical Methods in Natural Language Processing (EMNLP)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing","raw_type":"proceedings-article"},"volume":null,"type":"conference-paper","is_traditional":true,"indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"bronze","oa_url":"https://aclanthology.org/2026.emnlp-main.1549.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5157768581","display_name":"Luoyu Chen","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I114017466","display_name":"University of Technology Sydney","ror":"https://ror.org/03f0f6041","country_code":"AU","type":"education","lineage":["https://openalex.org/I114017466"]}],"countries":["AU"],"is_corresponding":false,"raw_author_name":"Luoyu Chen","raw_affiliation_strings":["University of Technology Sydney , Sydney , NSW , Australia"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Technology Sydney , Sydney , NSW , Australia","institution_ids":["https://openalex.org/I114017466"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5157676619","display_name":"Weiqi Wang","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I87445476","display_name":"Xi'an Jiaotong University","ror":"https://ror.org/017zhmm22","country_code":"CN","type":"education","lineage":["https://openalex.org/I87445476"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Weiqi Wang","raw_affiliation_strings":["Xi'an Jiaotong University , Xi'an , Shaanxi , China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Xi'an Jiaotong University , Xi'an , Shaanxi , China","institution_ids":["https://openalex.org/I87445476"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5059567156","display_name":"Zhiyi Tian","orcid":"https://orcid.org/0000-0001-8905-0941","observed_orcids":["https://orcid.org/0000-0001-8905-0941"]},"institutions":[{"id":"https://openalex.org/I76569877","display_name":"Southeast University","ror":"https://ror.org/04ct4d772","country_code":"CN","type":"education","lineage":["https://openalex.org/I76569877"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhiyi Tian","raw_affiliation_strings":["Southeast University , Nanjing , Jiangsu , China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Southeast University , Nanjing , Jiangsu , China","institution_ids":["https://openalex.org/I76569877"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5160095104","display_name":"Chenhan Zhang","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I114017466","display_name":"University of Technology Sydney","ror":"https://ror.org/03f0f6041","country_code":"AU","type":"education","lineage":["https://openalex.org/I114017466"]}],"countries":["AU"],"is_corresponding":false,"raw_author_name":"Chenhan Zhang","raw_affiliation_strings":["University of Technology Sydney , Sydney , NSW , Australia"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Technology Sydney , Sydney , NSW , Australia","institution_ids":["https://openalex.org/I114017466"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5157664721","display_name":"Shui Yu","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I114017466","display_name":"University of Technology Sydney","ror":"https://ror.org/03f0f6041","country_code":"AU","type":"education","lineage":["https://openalex.org/I114017466"]}],"countries":["AU"],"is_corresponding":false,"raw_author_name":"Shui Yu","raw_affiliation_strings":["University of Technology Sydney , Sydney , NSW , Australia"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Technology Sydney , Sydney , NSW , Australia","institution_ids":["https://openalex.org/I114017466"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":3,"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":0.67142474,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"35342","last_page":"35366"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11689","display_name":"Adversarial Robustness in Machine Learning","score":1.0,"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":1.0,"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":0.0,"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":0.0,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"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/jailbreak-attacks","display_name":"jailbreak attacks","score":0.8790000081062317},{"id":"https://openalex.org/keywords/intent-extraction","display_name":"intent extraction","score":0.7459999918937683},{"id":"https://openalex.org/keywords/over-refusal","display_name":"over-refusal","score":0.8420000076293945},{"id":"https://openalex.org/keywords/adversarial-prompts","display_name":"adversarial prompts","score":0.6340000033378601},{"id":"https://openalex.org/keywords/harm-mitigation","display_name":"harm mitigation","score":0.3149999976158142}],"concepts":[{"id":"https://openalex.org/C37736160","wikidata":"https://www.wikidata.org/wiki/Q1801315","display_name":"Adversarial system","level":2,"score":0.7509999871253967},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.6480000019073486},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.6420999765396118},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.5726000070571899},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5634999871253967},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.5133000016212463},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.45809999108314514},{"id":"https://openalex.org/C2777363581","wikidata":"https://www.wikidata.org/wiki/Q15098235","display_name":"Harm","level":2,"score":0.3758000135421753},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.34940001368522644},{"id":"https://openalex.org/C112930515","wikidata":"https://www.wikidata.org/wiki/Q4389547","display_name":"Risk analysis (engineering)","level":1,"score":0.27619999647140503},{"id":"https://openalex.org/C16678853","wikidata":"https://www.wikidata.org/wiki/Q486972","display_name":"Human settlement","level":2,"score":0.27480000257492065},{"id":"https://openalex.org/C189950617","wikidata":"https://www.wikidata.org/wiki/Q937228","display_name":"Property (philosophy)","level":2,"score":0.2696000039577484},{"id":"https://openalex.org/C108827166","wikidata":"https://www.wikidata.org/wiki/Q175975","display_name":"Internet privacy","level":1,"score":0.26260000467300415}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.18653/2026.emnlp-main.1549","doi":"https://doi.org/10.18653/2026.emnlp-main.1549","is_oa":true,"landing_page_url":"https://doi.org/10.18653/2026.emnlp-main.1549","pdf_url":"https://aclanthology.org/2026.emnlp-main.1549.pdf","source":{"id":"https://openalex.org/S4363608991","display_name":"Conference on Empirical Methods in Natural Language Processing (EMNLP)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2610.04470","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2610.04470","pdf_url":"https://arxiv.org/pdf/2610.04470","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2610.04470","doi":"https://doi.org/10.48550/arxiv.2610.04470","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2610.04470","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.18653/2026.emnlp-main.1549","doi":"https://doi.org/10.18653/2026.emnlp-main.1549","is_oa":true,"landing_page_url":"https://doi.org/10.18653/2026.emnlp-main.1549","pdf_url":"https://aclanthology.org/2026.emnlp-main.1549.pdf","source":{"id":"https://openalex.org/S4363608991","display_name":"Conference on Empirical Methods in Natural Language Processing (EMNLP)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing","raw_type":"proceedings-article"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7220686358.pdf","grobid_xml":"https://content.openalex.org/works/W7220686358.grobid-xml"},"referenced_works_count":17,"referenced_works":["https://openalex.org/W2794325560","https://openalex.org/W3210904070","https://openalex.org/W4322718191","https://openalex.org/W4327810158","https://openalex.org/W4385374425","https://openalex.org/W4385894687","https://openalex.org/W4387436631","https://openalex.org/W4387559087","https://openalex.org/W4387561528","https://openalex.org/W4392182141","https://openalex.org/W4392222150","https://openalex.org/W4399317891","https://openalex.org/W4402671676","https://openalex.org/W4403322022","https://openalex.org/W4403590208","https://openalex.org/W4410609100","https://openalex.org/W4415797762"],"related_works":["https://openalex.org/W4390897726","https://openalex.org/W7076365198","https://openalex.org/W4407425033","https://openalex.org/W4416618290","https://openalex.org/W4414602076","https://openalex.org/W7166782477","https://openalex.org/W7083823696","https://openalex.org/W4414360535","https://openalex.org/W6966788169","https://openalex.org/W7214285256"],"abstract_inverted_index":{"Large":[0],"language":[1],"models":[2],"(LLMs)":[3],"remain":[4],"vulnerable":[5],"to":[6,74,90,100,120,122,131,147],"jailbreak":[7,46,69,117,141],"attacks":[8,133],"that":[9,48,59,114],"conceal":[10],"harmful":[11],"intent":[12,32,53],"within":[13],"complex":[14],"adversarial":[15],"prompts.Existing":[16],"defenses":[17],"primarily":[18],"rely":[19],"on":[20,108],"input":[21],"perturbation":[22],"or":[23],"harmful-output":[24],"suppression,":[25],"but":[26],"they":[27],"rarely":[28],"model":[29],"where":[30],"malicious":[31],"resides,":[33],"resulting":[34],"in":[35],"brittle":[36],"protection":[37],"and":[38,103,134],"excessive":[39],"overrefusal.We":[40],"propose":[41],"SENTINEL,":[42],"a":[43,136],"plug-and-play,":[44],"generation-time":[45],"defense":[47],"reframes":[49],"mitigation":[50],"as":[51],"an":[52],"extraction":[54],"problem.Our":[55],"key":[56],"insight":[57],"is":[58],"instruction-tuned":[60],"LLMs":[61,112],"exhibit":[62],"strong":[63],"input-output":[64],"semantic":[65],"consistency:":[66],"regardless":[67],"of":[68],"complexity,":[70],"generated":[71],"outputs":[72],"tend":[73],"align":[75],"with":[76],"the":[77],"attacker's":[78],"true":[79],"intent.SENTINEL":[80],"exploits":[81],"this":[82],"property":[83],"by":[84],"matching":[85],"semantically":[86],"aligned":[87,148],"inputoutput":[88],"regions":[89],"extract":[91],"intention-revealing":[92],"subsequences,":[93],"scores":[94],"these":[95],"subsequences":[96],"using":[97],"refusal-direction":[98],"projections":[99],"estimate":[101],"harmfulness,":[102],"halts":[104],"generation":[105],"when":[106],"necessary.Experiments":[107],"HarmBench":[109],"across":[110],"multiple":[111],"show":[113],"SENTINEL":[115,139],"reduces":[116],"success":[118],"rates":[119],"close":[121],"5%":[123],"while":[124],"maintaining":[125],"low":[126],"over-refusal.We":[127],"further":[128],"demonstrate":[129],"robustness":[130],"adaptive":[132],"provide":[135],"mechanistic":[137],"interpretation:":[138],"re-distributes":[140],"features":[142],"from":[143],"alignment":[144],"blind":[145],"spots":[146],"regions.":[149]},"counts_by_year":[],"updated_date":"2026-11-08T12:12:28.789561","created_date":"2026-10-07T00:00:00"}