{"id":"https://openalex.org/W4412888670","doi":"https://doi.org/10.18653/v1/2025.findings-acl.332","title":"Beyond Reactive Safety: Risk-Aware LLM Alignment via Long-Horizon Simulation","display_name":"Beyond Reactive Safety: Risk-Aware LLM Alignment via Long-Horizon Simulation","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4412888670","doi":"https://doi.org/10.18653/v1/2025.findings-acl.332"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.findings-acl.332","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.332","pdf_url":"https://aclanthology.org/2025.findings-acl.332.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.332.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5120315379","display_name":"Chuanbin Sun","orcid":"https://orcid.org/0000-0003-2076-1133"},"institutions":[{"id":"https://openalex.org/I157725225","display_name":"University of Illinois Urbana-Champaign","ror":"https://ror.org/047426m28","country_code":"US","type":"education","lineage":["https://openalex.org/I157725225"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Chenkai Sun","raw_affiliation_strings":["University of Illinois Urbana-Champaign"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Illinois Urbana-Champaign","institution_ids":["https://openalex.org/I157725225"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100366431","display_name":"Di Zhang","orcid":"https://orcid.org/0000-0003-2526-8954"},"institutions":[{"id":"https://openalex.org/I157725225","display_name":"University of Illinois Urbana-Champaign","ror":"https://ror.org/047426m28","country_code":"US","type":"education","lineage":["https://openalex.org/I157725225"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Denghui Zhang","raw_affiliation_strings":["University of Illinois Urbana-Champaign"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Illinois Urbana-Champaign","institution_ids":["https://openalex.org/I157725225"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5028518494","display_name":"ChengXiang Zhai","orcid":"https://orcid.org/0000-0002-6434-3702"},"institutions":[{"id":"https://openalex.org/I157725225","display_name":"University of Illinois Urbana-Champaign","ror":"https://ror.org/047426m28","country_code":"US","type":"education","lineage":["https://openalex.org/I157725225"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"ChengXiang Zhai","raw_affiliation_strings":["University of Illinois Urbana-Champaign"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Illinois Urbana-Champaign","institution_ids":["https://openalex.org/I157725225"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5075033889","display_name":"Heng Ji","orcid":"https://orcid.org/0000-0002-0464-7966"},"institutions":[{"id":"https://openalex.org/I157725225","display_name":"University of Illinois Urbana-Champaign","ror":"https://ror.org/047426m28","country_code":"US","type":"education","lineage":["https://openalex.org/I157725225"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Heng Ji","raw_affiliation_strings":["University of Illinois Urbana-Champaign"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Illinois Urbana-Champaign","institution_ids":["https://openalex.org/I157725225"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I157725225"],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":1,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"6422","last_page":"6434"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T13295","display_name":"Safety Systems Engineering in Autonomy","score":0.9424999952316284,"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"}},"topics":[{"id":"https://openalex.org/T13295","display_name":"Safety Systems Engineering in Autonomy","score":0.9424999952316284,"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"}},{"id":"https://openalex.org/T10703","display_name":"Business Process Modeling and Analysis","score":0.9368000030517578,"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/computer-science","display_name":"Computer science","score":0.6213019490242004},{"id":"https://openalex.org/keywords/horizon","display_name":"Horizon","score":0.5159834027290344},{"id":"https://openalex.org/keywords/risk-analysis","display_name":"Risk analysis (engineering)","score":0.3781861364841461},{"id":"https://openalex.org/keywords/medicine","display_name":"Medicine","score":0.1386423408985138},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.10297256708145142}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6213019490242004},{"id":"https://openalex.org/C159176650","wikidata":"https://www.wikidata.org/wiki/Q43261","display_name":"Horizon","level":2,"score":0.5159834027290344},{"id":"https://openalex.org/C112930515","wikidata":"https://www.wikidata.org/wiki/Q4389547","display_name":"Risk analysis (engineering)","level":1,"score":0.3781861364841461},{"id":"https://openalex.org/C71924100","wikidata":"https://www.wikidata.org/wiki/Q11190","display_name":"Medicine","level":0,"score":0.1386423408985138},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.10297256708145142},{"id":"https://openalex.org/C2524010","wikidata":"https://www.wikidata.org/wiki/Q8087","display_name":"Geometry","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.findings-acl.332","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.332","pdf_url":"https://aclanthology.org/2025.findings-acl.332.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.332","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.332","pdf_url":"https://aclanthology.org/2025.findings-acl.332.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":[{"id":"https://metadata.un.org/sdg/13","display_name":"Climate action","score":0.7599999904632568}],"awards":[{"id":"https://openalex.org/G7091742882","display_name":null,"funder_award_id":"HR001121C0165","funder_id":"https://openalex.org/F4320332180","funder_display_name":"Defense Advanced Research Projects Agency"},{"id":"https://openalex.org/G7559374616","display_name":null,"funder_award_id":"FA8650-23-C-7316","funder_id":"https://openalex.org/F4320332180","funder_display_name":"Defense Advanced Research Projects Agency"}],"funders":[{"id":"https://openalex.org/F4320332180","display_name":"Defense Advanced Research Projects Agency","ror":"https://ror.org/02caytj08"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4412888670.pdf","grobid_xml":"https://content.openalex.org/works/W4412888670.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W2390279801","https://openalex.org/W4391913857","https://openalex.org/W2358668433","https://openalex.org/W4396701345","https://openalex.org/W2376932109","https://openalex.org/W2001405890","https://openalex.org/W4396696052"],"abstract_inverted_index":{"Given":[0],"the":[1,23,54,92],"growing":[2],"influence":[3],"of":[4,26,58,66],"language":[5,59],"modelbased":[6],"agents":[7],"on":[8,43,91,106],"high-stakes":[9],"societal":[10,41],"decisions,":[11],"from":[12,79],"public":[13],"policy":[14],"to":[15,74],"healthcare,":[16],"ensuring":[17],"their":[18,27],"beneficial":[19],"impact":[20],"requires":[21],"understanding":[22],"far-reaching":[24],"implications":[25],"suggestions.We":[28],"propose":[29],"a":[30,44,64,114],"proof-of-concept":[31],"framework":[32],"that":[33],"projects":[34],"how":[35],"model-generated":[36],"advice":[37],"could":[38],"propagate":[39],"through":[40],"systems":[42],"macroscopic":[45],"scale":[46],"over":[47,88],"time,":[48],"enabling":[49],"more":[50],"robust":[51],"alignment.To":[52],"assess":[53],"long-term":[55],"safety":[56,108],"awareness":[57],"models,":[60],"we":[61],"also":[62,96],"introduce":[63],"dataset":[65,94],"100":[67],"indirect":[68],"harm":[69],"scenarios,":[70],"testing":[71],"models'":[72],"ability":[73],"foresee":[75],"adverse,":[76],"nonobvious":[77],"outcomes":[78],"seemingly":[80],"harmless":[81],"user":[82],"prompts.Our":[83],"approach":[84],"achieves":[85],"not":[86],"only":[87],"20%":[89],"improvement":[90],"new":[93],"but":[95],"an":[97],"average":[98],"win":[99],"rate":[100],"exceeding":[101],"70%":[102],"against":[103],"strong":[104],"baselines":[105],"existing":[107],"benchmarks":[109],"(AdvBench,":[110],"SafeRLHF,":[111],"WildGuardMix),":[112],"suggesting":[113],"promising":[115],"direction":[116],"for":[117],"safer":[118],"agents.":[119]},"counts_by_year":[{"year":2026,"cited_by_count":1}],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}