{"id":"https://openalex.org/W7166874769","doi":"https://doi.org/10.18653/v1/2026.acl-long.1570","title":"Resolving the Security-Auditability Dilemma with Auditable Latent Chain-of-Thought Alignment","display_name":"Resolving the Security-Auditability Dilemma with Auditable Latent Chain-of-Thought Alignment","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7166874769","doi":"https://doi.org/10.18653/v1/2026.acl-long.1570"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.acl-long.1570","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.1570","pdf_url":"https://aclanthology.org/2026.acl-long.1570.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.1570.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5139842138","display_name":"Guan Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Guan Wang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5064648779","display_name":"Biyu Zhou","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Biyu Zhou","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5047580108","display_name":"Xuehai Tang","orcid":"https://orcid.org/0000-0002-9901-4087"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xuehai Tang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5112539471","display_name":"Jizhong Han","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jizhong Han","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5139848109","display_name":"Songlin Hu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Songlin Hu","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":0.91062923,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"34051","last_page":"34067"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10927","display_name":"Access Control and Trust","score":0.11949999630451202,"subfield":{"id":"https://openalex.org/subfields/3312","display_name":"Sociology and Political Science"},"field":{"id":"https://openalex.org/fields/33","display_name":"Social Sciences"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},"topics":[{"id":"https://openalex.org/T10927","display_name":"Access Control and Trust","score":0.11949999630451202,"subfield":{"id":"https://openalex.org/subfields/3312","display_name":"Sociology and Political Science"},"field":{"id":"https://openalex.org/fields/33","display_name":"Social Sciences"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},{"id":"https://openalex.org/T10639","display_name":"Advanced Software Engineering Methodologies","score":0.09780000150203705,"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/T12127","display_name":"Software System Performance and Reliability","score":0.08789999783039093,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"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/dilemma","display_name":"Dilemma","score":0.5468999743461609},{"id":"https://openalex.org/keywords/action","display_name":"Action (physics)","score":0.32350000739097595},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.2896000146865845},{"id":"https://openalex.org/keywords/control","display_name":"Control (management)","score":0.28369998931884766},{"id":"https://openalex.org/keywords/identification","display_name":"Identification (biology)","score":0.2232999950647354}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5835000276565552},{"id":"https://openalex.org/C2778496695","wikidata":"https://www.wikidata.org/wiki/Q254128","display_name":"Dilemma","level":2,"score":0.5468999743461609},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3702999949455261},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.32350000739097595},{"id":"https://openalex.org/C112930515","wikidata":"https://www.wikidata.org/wiki/Q4389547","display_name":"Risk analysis (engineering)","level":1,"score":0.3230000138282776},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.2896000146865845},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.28790000081062317},{"id":"https://openalex.org/C2775924081","wikidata":"https://www.wikidata.org/wiki/Q55608371","display_name":"Control (management)","level":2,"score":0.28369998931884766},{"id":"https://openalex.org/C116834253","wikidata":"https://www.wikidata.org/wiki/Q2039217","display_name":"Identification (biology)","level":2,"score":0.2232999950647354},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.2134000062942505}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2026.acl-long.1570","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.1570","pdf_url":"https://aclanthology.org/2026.acl-long.1570.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.1570","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.1570","pdf_url":"https://aclanthology.org/2026.acl-long.1570.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/G5208670511","display_name":null,"funder_award_id":"U24A20335","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"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7166874769.pdf","grobid_xml":"https://content.openalex.org/works/W7166874769.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"To":[0],"address":[1,94],"the":[2,20,27,55,82,113,123,129,136,161,184],"increasingly":[3],"severe":[4],"safety":[5,13,34,114,124],"risk":[6],"of":[7,22,33,131,187],"large":[8],"language":[9],"models":[10],"(LLMs),":[11],"reasoningbased":[12],"alignment":[14],"methods":[15,18],"have":[16],"emerged.These":[17],"overcome":[19],"limitations":[21],"'shallow":[23],"alignment'":[24],"by":[25,191],"exposing":[26],"model's":[28],"Chain-of-Thought":[29],"(CoT),":[30],"enabling":[31],"auditability":[32],"reasoning":[35,61,92,108,125,167],"process":[36,116,126,148],"through":[37],"both":[38,210],"training-phase":[39],"supervision":[40,172],"and":[41,80,144,213],"post-generation":[42],"verification.However,":[43],"this":[44,147],"transparency":[45],"creates":[46],"a":[47,50,63,103,118,151,155,203],"critical":[48],"vulnerability,":[49],"tension":[51],"we":[52,96],"define":[53],"as":[54],"Security":[56],"Auditability":[57],"Dilemma:":[58],"while":[59,134,198],"explicit":[60],"is":[62,149],"prerequisite":[64],"for":[65,77,171],"safety,":[66],"its":[67,90,165],"textual":[68,138],"Auditable":[69,98],"paradoxically":[70],"transforms":[71],"it":[72],"into":[73,117,168],"an":[74],"optimization":[75],"target":[76],"adaptive":[78,145,188],"attackers":[79],"induces":[81],"model":[83,162],"to":[84,127,163,195],"unintentionally":[85],"copy":[86],"harmful":[87],"content":[88],"from":[89,109],"own":[91],"context.To":[93],"this,":[95],"propose":[97],"Latent":[99],"CoT":[100],"Alignment":[101],"(ALCA),":[102],"framework":[104,201],"that":[105,140,159,178,208],"decouples":[106],"internal":[107,142],"external":[110],"output.ALCA":[111],"shifts":[112],"deliberation":[115],"continuous":[119],"latent":[120,166],"space.This":[121],"allows":[122,160],"guide":[128],"generation":[130],"harmless":[132],"outputs,":[133],"eliminates":[135],"discrete":[137],"surface":[139],"facilitates":[141],"copying":[143],"attack.Yet,":[146],"not":[150],"black":[152],"box.we":[153],"introduce":[154],"restricted":[156],"Self-Decoding":[157],"mechanism":[158],"reconstruct":[164],"human-readable":[169],"text":[170],"under":[173],"specific":[174],"guidance.Extensive":[175],"experiments":[176],"show":[177],"ALCA":[179],"achieves":[180],"robustness":[181],"alignment,":[182],"reducing":[183],"success":[185],"rate":[186],"jailbreak":[189],"attacks":[190],"over":[192],"40%":[193],"compared":[194],"strong":[196],"baselines,":[197],"preserving":[199],"performance.Our":[200],"presents":[202],"path":[204],"toward":[205],"building":[206],"LLMs":[207],"are":[209],"robustly":[211],"secure":[212],"auditable.":[214]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2026-07-02T00:00:00"}
