{"id":"https://openalex.org/W7166878564","doi":"https://doi.org/10.18653/v1/2026.findings-acl.1390","title":"LCO: LLM-based Constraint Optimization for Safer Agentic LLMs in Real-world Tasks","display_name":"LCO: LLM-based Constraint Optimization for Safer Agentic LLMs in Real-world Tasks","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7166878564","doi":"https://doi.org/10.18653/v1/2026.findings-acl.1390"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.findings-acl.1390","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1390","pdf_url":"https://aclanthology.org/2026.findings-acl.1390.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 2026","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2026.findings-acl.1390.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5139717284","display_name":"Jiayong Wan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jiayong Wan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139756132","display_name":"Jiawei Chen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jiawei Chen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139837553","display_name":"Zhaoxia Yin","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhaoxia Yin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138492788","display_name":"Shuyuan Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu Shuyuan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5139812916","display_name":"Hang Su","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hang Su","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":0.86819396,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"27910","last_page":"27934"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11596","display_name":"Constraint Satisfaction and Optimization","score":0.3962000012397766,"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"}},"topics":[{"id":"https://openalex.org/T11596","display_name":"Constraint Satisfaction and Optimization","score":0.3962000012397766,"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"}},{"id":"https://openalex.org/T10456","display_name":"Multi-Agent Systems and Negotiation","score":0.1678999960422516,"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/T10906","display_name":"AI-based Problem Solving and Planning","score":0.03709999844431877,"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/safer","display_name":"SAFER","score":0.5776000022888184},{"id":"https://openalex.org/keywords/constraint","display_name":"Constraint (computer-aided design)","score":0.4494999945163727},{"id":"https://openalex.org/keywords/action","display_name":"Action (physics)","score":0.3248000144958496},{"id":"https://openalex.org/keywords/minification","display_name":"Minification","score":0.3181000053882599},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.289000004529953},{"id":"https://openalex.org/keywords/constrained-optimization","display_name":"Constrained optimization","score":0.287200003862381}],"concepts":[{"id":"https://openalex.org/C2776654903","wikidata":"https://www.wikidata.org/wiki/Q2601463","display_name":"SAFER","level":2,"score":0.5776000022888184},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.48829999566078186},{"id":"https://openalex.org/C2776036281","wikidata":"https://www.wikidata.org/wiki/Q48769818","display_name":"Constraint (computer-aided design)","level":2,"score":0.4494999945163727},{"id":"https://openalex.org/C112930515","wikidata":"https://www.wikidata.org/wiki/Q4389547","display_name":"Risk analysis (engineering)","level":1,"score":0.4422000050544739},{"id":"https://openalex.org/C162324750","wikidata":"https://www.wikidata.org/wiki/Q8134","display_name":"Economics","level":0,"score":0.3497999906539917},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.3248000144958496},{"id":"https://openalex.org/C147764199","wikidata":"https://www.wikidata.org/wiki/Q6865248","display_name":"Minification","level":2,"score":0.3181000053882599},{"id":"https://openalex.org/C144133560","wikidata":"https://www.wikidata.org/wiki/Q4830453","display_name":"Business","level":0,"score":0.29030001163482666},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.289000004529953},{"id":"https://openalex.org/C55660270","wikidata":"https://www.wikidata.org/wiki/Q5164377","display_name":"Constrained optimization","level":2,"score":0.287200003862381},{"id":"https://openalex.org/C18762648","wikidata":"https://www.wikidata.org/wiki/Q42213","display_name":"Work (physics)","level":2,"score":0.2671000063419342},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.2581999897956848},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.2554999887943268},{"id":"https://openalex.org/C64543145","wikidata":"https://www.wikidata.org/wiki/Q162942","display_name":"Intersection (aeronautics)","level":2,"score":0.25360000133514404},{"id":"https://openalex.org/C539667460","wikidata":"https://www.wikidata.org/wiki/Q2414942","display_name":"Management science","level":1,"score":0.2533999979496002}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2026.findings-acl.1390","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1390","pdf_url":"https://aclanthology.org/2026.findings-acl.1390.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 2026","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.findings-acl.1390","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1390","pdf_url":"https://aclanthology.org/2026.findings-acl.1390.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 2026","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G1855097118","display_name":null,"funder_award_id":"ZR2022ZD01","funder_id":"https://openalex.org/F4320324174","funder_display_name":"Natural Science Foundation of Shandong Province"},{"id":"https://openalex.org/G387170554","display_name":null,"funder_award_id":"ZR2022ZD01","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G7681595199","display_name":null,"funder_award_id":"62472177","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"},{"id":"https://openalex.org/F4320324174","display_name":"Natural Science Foundation of Shandong Province","ror":null}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7166878564.pdf","grobid_xml":"https://content.openalex.org/works/W7166878564.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Large":[0],"Language":[1],"Models":[2],"(LLMs)":[3],"are":[4,44],"increasingly":[5],"acting":[6],"as":[7,50],"autonomous":[8],"agents,":[9],"but":[10,57],"their":[11,31],"continuous":[12],"interaction":[13],"with":[14],"the":[15,59,89,114,141,152,161,167],"environment":[16],"can":[17],"lead":[18],"to":[19,33,46,91,112],"in-context":[20],"reward":[21],"hacking":[22],"(ICRH),":[23],"a":[24,72,118,148],"phenomenon":[25],"in":[26,133,151],"which":[27,87,106],"LLMs":[28],"iteratively":[29],"optimize":[30],"behavior":[32],"maximize":[34],"proxy":[35],"objectives,":[36],"inadvertently":[37],"producing":[38],"harmful":[39],"side":[40],"effects.Existing":[41],"defense":[42],"methods":[43],"insufficient":[45],"address":[47],"this":[48,64],"risk,":[49],"ICRH":[51,77,132,168],"arises":[52],"not":[53],"from":[54,58],"adversarial":[55],"inputs":[56],"model's":[60,115],"own":[61],"over-optimization.To":[62],"mitigate":[63],"issue,":[65],"we":[66],"propose":[67],"LLM-based":[68,108],"Constraint":[69],"Optimization":[70],"(LCO),":[71],"framework":[73],"that":[74,128],"effectively":[75],"reduces":[76,166],"without":[78,176],"model":[79],"fine-tuning.LCO":[80],"consists":[81],"of":[82],"two":[83],"modules:":[84],"self-thought":[85],"module,":[86,105],"guides":[88],"LLM":[90],"proactively":[92],"deliberate":[93],"and":[94,101,110,136],"integrate":[95],"potential":[96],"safety":[97,174],"constraints":[98],"before":[99],"execution;":[100],"guided":[102],"evolutionary":[103],"exploration":[104],"employs":[107],"crossover":[109],"mutation":[111],"constrain":[113],"actions":[116],"within":[117],"safe":[119],"solution":[120],"space":[121],"while":[122,159],"maintaining":[123],"task":[124,178],"performance.Experimental":[125],"results":[126],"demonstrate":[127],"LCO":[129,146],"substantially":[130],"alleviates":[131],"both":[134],"output-refine":[135],"policy-refine":[137],"scenarios.In":[138],"particular,":[139],"on":[140,157,160],"tweet":[142],"engagement":[143],"optimization":[144,163],"task,":[145],"achieves":[147],"39%":[149],"reduction":[150],"Toxicity":[153],"Growth":[154],"Rate":[155,170],"(TGR)":[156],"GPT-4,":[158],"policy":[162],"benchmark,":[164],"it":[165],"Occurrence":[169],"by":[171],"15.23%,":[172],"demonstrating":[173],"improvement":[175],"sacrificing":[177],"performance.Our":[179],"code":[180],"is":[181],"available":[182],"at:":[183],"https://github.com/":[184],"Califoni/LCO_for_ICRH.":[185]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2026-07-02T00:00:00"}