{"id":"https://openalex.org/W7138016746","doi":"https://doi.org/10.1609/aaai.v40i2.37106","title":"Where to Start Alignment? Diffusion Large Language Model May Demand a Distinct Position","display_name":"Where to Start Alignment? Diffusion Large Language Model May Demand a Distinct Position","publication_year":2026,"publication_date":"2026-03-14","ids":{"openalex":"https://openalex.org/W7138016746","doi":"https://doi.org/10.1609/aaai.v40i2.37106"},"language":null,"primary_location":{"id":"doi:10.1609/aaai.v40i2.37106","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aaai.v40i2.37106","pdf_url":"https://ojs.aaai.org/index.php/AAAI/article/download/37106/41068","source":{"id":"https://openalex.org/S4210191458","display_name":"Proceedings of the AAAI Conference on Artificial Intelligence","issn_l":"2159-5399","issn":["2159-5399","2374-3468"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/P4310320058","host_organization_name":"Association for the Advancement of Artificial Intelligence","host_organization_lineage":["https://openalex.org/P4310320058"],"host_organization_lineage_names":["Association for the Advancement of Artificial Intelligence"],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the AAAI Conference on Artificial Intelligence","raw_type":"journal-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"diamond","oa_url":"https://ojs.aaai.org/index.php/AAAI/article/download/37106/41068","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5123525257","display_name":"Zhixin Xie","orcid":null},"institutions":[{"id":"https://openalex.org/I172675005","display_name":"Nanyang Technological University","ror":"https://ror.org/02e7b5302","country_code":"SG","type":"education","lineage":["https://openalex.org/I172675005"]}],"countries":["SG"],"is_corresponding":false,"raw_author_name":"Zhixin Xie","raw_affiliation_strings":["Nanyang Technological University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Nanyang Technological University","institution_ids":["https://openalex.org/I172675005"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129670589","display_name":"Xurui Song","orcid":null},"institutions":[{"id":"https://openalex.org/I172675005","display_name":"Nanyang Technological University","ror":"https://ror.org/02e7b5302","country_code":"SG","type":"education","lineage":["https://openalex.org/I172675005"]}],"countries":["SG"],"is_corresponding":false,"raw_author_name":"Xurui Song","raw_affiliation_strings":["Nanyang Technological University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Nanyang Technological University","institution_ids":["https://openalex.org/I172675005"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5129647722","display_name":"Jun Luo","orcid":null},"institutions":[{"id":"https://openalex.org/I172675005","display_name":"Nanyang Technological University","ror":"https://ror.org/02e7b5302","country_code":"SG","type":"education","lineage":["https://openalex.org/I172675005"]}],"countries":["SG"],"is_corresponding":false,"raw_author_name":"Jun Luo","raw_affiliation_strings":["Nanyang Technological University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Nanyang Technological University","institution_ids":["https://openalex.org/I172675005"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I172675005"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.2129981,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"40","issue":"2","first_page":"1328","last_page":"1336"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.5831000208854675,"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":0.5831000208854675,"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":0.14710000157356262,"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/T12026","display_name":"Explainable Artificial Intelligence (XAI)","score":0.039799999445676804,"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/position","display_name":"Position (finance)","score":0.5530999898910522},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.508899986743927},{"id":"https://openalex.org/keywords/forcing","display_name":"Forcing (mathematics)","score":0.4361000061035156},{"id":"https://openalex.org/keywords/power","display_name":"Power (physics)","score":0.3260999917984009},{"id":"https://openalex.org/keywords/order","display_name":"Order (exchange)","score":0.3077000081539154},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.3068999946117401}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6861000061035156},{"id":"https://openalex.org/C198082294","wikidata":"https://www.wikidata.org/wiki/Q3399648","display_name":"Position (finance)","level":2,"score":0.5530999898910522},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.508899986743927},{"id":"https://openalex.org/C197115733","wikidata":"https://www.wikidata.org/wiki/Q1003136","display_name":"Forcing (mathematics)","level":2,"score":0.4361000061035156},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.3481999933719635},{"id":"https://openalex.org/C163258240","wikidata":"https://www.wikidata.org/wiki/Q25342","display_name":"Power (physics)","level":2,"score":0.3260999917984009},{"id":"https://openalex.org/C182306322","wikidata":"https://www.wikidata.org/wiki/Q1779371","display_name":"Order (exchange)","level":2,"score":0.3077000081539154},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.3068999946117401},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.29670000076293945},{"id":"https://openalex.org/C192209626","wikidata":"https://www.wikidata.org/wiki/Q190909","display_name":"Focus (optics)","level":2,"score":0.28999999165534973},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.25600001215934753},{"id":"https://openalex.org/C112972136","wikidata":"https://www.wikidata.org/wiki/Q7595718","display_name":"Stability (learning theory)","level":2,"score":0.2540999948978424}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1609/aaai.v40i2.37106","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aaai.v40i2.37106","pdf_url":"https://ojs.aaai.org/index.php/AAAI/article/download/37106/41068","source":{"id":"https://openalex.org/S4210191458","display_name":"Proceedings of the AAAI Conference on Artificial Intelligence","issn_l":"2159-5399","issn":["2159-5399","2374-3468"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/P4310320058","host_organization_name":"Association for the Advancement of Artificial Intelligence","host_organization_lineage":["https://openalex.org/P4310320058"],"host_organization_lineage_names":["Association for the Advancement of Artificial Intelligence"],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the AAAI Conference on Artificial Intelligence","raw_type":"journal-article"}],"best_oa_location":{"id":"doi:10.1609/aaai.v40i2.37106","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aaai.v40i2.37106","pdf_url":"https://ojs.aaai.org/index.php/AAAI/article/download/37106/41068","source":{"id":"https://openalex.org/S4210191458","display_name":"Proceedings of the AAAI Conference on Artificial Intelligence","issn_l":"2159-5399","issn":["2159-5399","2374-3468"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/P4310320058","host_organization_name":"Association for the Advancement of Artificial Intelligence","host_organization_lineage":["https://openalex.org/P4310320058"],"host_organization_lineage_names":["Association for the Advancement of Artificial Intelligence"],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the AAAI Conference on Artificial Intelligence","raw_type":"journal-article"},"sustainable_development_goals":[{"score":0.5184822678565979,"id":"https://metadata.un.org/sdg/4","display_name":"Quality Education"}],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7138016746.pdf","grobid_xml":"https://content.openalex.org/works/W7138016746.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Diffusion":[0],"Large":[1],"Language":[2],"Models":[3],"(dLLMs)":[4],"have":[5,122,133],"recently":[6],"emerged":[7],"as":[8,129],"a":[9,25,48,62,134,138,170],"competitive":[10],"non-autoregressive":[11],"paradigm":[12],"due":[13],"to":[14,54,94,103,113,125,147],"their":[15,55],"unique":[16,56],"training":[17],"and":[18,46,68,151,189,212],"inference":[19],"approach.":[20],"However,":[21],"there":[22],"is":[23],"currently":[24],"lack":[26],"of":[27,42,72,83,98,206,221],"safety":[28,44,50,97],"study":[29],"on":[30,118,161,198,209],"this":[31,35,101,149,162],"novel":[32,49,171],"architecture.":[33],"In":[34],"paper,":[36],"we":[37,60,77,130,164],"present":[38],"the":[39,66,75,80,84,88,95,114,119,145,156,176,204,219],"first":[40],"analysis":[41],"dLLMs'":[43],"performance":[45,193],"propose":[47],"alignment":[51],"method":[52,172],"tailored":[53],"generation":[57,140,179],"characteristics.":[58],"Specifically,":[59],"identify":[61],"critical":[63,93,157],"asymmetry":[64],"between":[65],"defender":[67],"attacker":[69],"in":[70,142],"terms":[71],"security.":[73],"For":[74],"defender,":[76],"reveal":[78],"that":[79,105,173],"middle":[81,107,127,158,178],"tokens":[82,108],"response,":[85],"rather":[86],"than":[87],"initial":[89],"ones,":[90],"are":[91],"more":[92,111],"overall":[96],"dLLM":[99,208],"outputs;":[100],"seems":[102],"suggest":[104],"aligning":[106],"can":[109],"be":[110],"beneficial":[112],"defender.":[115],"The":[116,215],"attacker,":[117],"contrary,":[120],"may":[121],"limited":[123],"power":[124],"manipulate":[126],"tokens,":[128],"find":[131],"dLLMs":[132],"strong":[135],"tendency":[136],"towards":[137],"sequential":[139],"order":[141],"practice,":[143],"forcing":[144],"attack":[146,196],"meet":[148],"distribution":[150],"diverting":[152],"it":[153],"from":[154],"influencing":[155],"tokens.":[159],"Building":[160],"asymmetry,":[163],"introduce":[165],"Middle-tOken":[166],"Safety":[167],"Alignment":[168],"(MOSA),":[169],"directly":[174],"aligns":[175],"model's":[177],"with":[180],"safe":[181],"refusals":[182],"exploiting":[183],"reinforcement":[184],"learning.":[185],"We":[186,201],"implement":[187],"MOSA":[188],"compare":[190],"its":[191],"security":[192],"against":[194],"eight":[195],"methods":[197],"two":[199],"benchmarks.":[200],"also":[202],"test":[203],"utility":[205],"MOSA-aligned":[207],"coding,":[210],"math,":[211],"general":[213],"reasoning.":[214],"results":[216],"strongly":[217],"prove":[218],"superiority":[220],"MOSA.":[222]},"counts_by_year":[],"updated_date":"2026-06-26T08:34:08.712188","created_date":"2026-03-18T00:00:00"}