{"id":"https://openalex.org/W4415230474","doi":"https://doi.org/10.1609/aies.v8i1.36541","title":"Evaluating Goal Drift in Language Model Agents","display_name":"Evaluating Goal Drift in Language Model Agents","publication_year":2025,"publication_date":"2025-10-15","ids":{"openalex":"https://openalex.org/W4415230474","doi":"https://doi.org/10.1609/aies.v8i1.36541"},"language":"en","primary_location":{"id":"doi:10.1609/aies.v8i1.36541","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aies.v8i1.36541","pdf_url":"https://ojs.aaai.org/index.php/AIES/article/download/36541/38679","source":{"id":"https://openalex.org/S5407048695","display_name":"Proceedings of the AAAI/ACM Conference on AI Ethics and Society","issn_l":"3065-8365","issn":["3065-8365"],"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society","raw_type":"journal-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"bronze","oa_url":"https://ojs.aaai.org/index.php/AIES/article/download/36541/38679","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5119937067","display_name":"Rauno Arike","orcid":null},"institutions":[{"id":"https://openalex.org/I179760745","display_name":"Mahaveer Academy of Technology and Science University","ror":"https://ror.org/0292bh813","country_code":"IN","type":"education","lineage":["https://openalex.org/I179760745"]}],"countries":["IN"],"is_corresponding":false,"raw_author_name":"Rauno Arike","raw_affiliation_strings":["ML Alignment & Theory Scholars (MATS)\nUniversity of Amsterdam"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"ML Alignment & Theory Scholars (MATS)\nUniversity of Amsterdam","institution_ids":["https://openalex.org/I179760745"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5119937068","display_name":"Elizabeth Donoway","orcid":null},"institutions":[{"id":"https://openalex.org/I95457486","display_name":"University of California, Berkeley","ror":"https://ror.org/01an7q238","country_code":"US","type":"education","lineage":["https://openalex.org/I95457486"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Elizabeth Donoway","raw_affiliation_strings":["ML Alignment & Theory Scholars (MATS)\nUniversity of California, Berkeley"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"ML Alignment & Theory Scholars (MATS)\nUniversity of California, Berkeley","institution_ids":["https://openalex.org/I95457486"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5012032714","display_name":"Henning Bartsch","orcid":null},"institutions":[{"id":"https://openalex.org/I4210088273","display_name":"Align Technology (United States)","ror":"https://ror.org/005sbaa41","country_code":"US","type":"company","lineage":["https://openalex.org/I4210088273"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Henning Bartsch","raw_affiliation_strings":["ML Alignment & Theory Scholars (MATS)"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"ML Alignment & Theory Scholars (MATS)","institution_ids":["https://openalex.org/I4210088273"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5034914617","display_name":"Marius Hobbhahn","orcid":"https://orcid.org/0009-0003-8244-3154"},"institutions":[{"id":"https://openalex.org/I4210154019","display_name":"Apollo Instruments (United States)","ror":"https://ror.org/04pmnt922","country_code":"US","type":"company","lineage":["https://openalex.org/I4210154019"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Marius Hobbhahn","raw_affiliation_strings":["Apollo Research"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Apollo Research","institution_ids":["https://openalex.org/I4210154019"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":4,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":8.3651,"has_fulltext":true,"cited_by_count":5,"citation_normalized_percentile":{"value":0.97935402,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":98,"max":99},"biblio":{"volume":"8","issue":"1","first_page":"192","last_page":"203"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10456","display_name":"Multi-Agent Systems and Negotiation","score":0.9854999780654907,"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/T10456","display_name":"Multi-Agent Systems and Negotiation","score":0.9854999780654907,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.942300021648407,"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/T12031","display_name":"Speech and dialogue systems","score":0.9217000007629395,"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/goal-orientation","display_name":"Goal orientation","score":0.5669999718666077},{"id":"https://openalex.org/keywords/context","display_name":"Context (archaeology)","score":0.5637999773025513},{"id":"https://openalex.org/keywords/goal-modeling","display_name":"Goal modeling","score":0.4553999900817871},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.4537000060081482},{"id":"https://openalex.org/keywords/goal-setting","display_name":"Goal setting","score":0.3343000113964081},{"id":"https://openalex.org/keywords/degree","display_name":"Degree (music)","score":0.32170000672340393}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6513000130653381},{"id":"https://openalex.org/C84653758","wikidata":"https://www.wikidata.org/wiki/Q5575175","display_name":"Goal orientation","level":2,"score":0.5669999718666077},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.5637999773025513},{"id":"https://openalex.org/C9586889","wikidata":"https://www.wikidata.org/wiki/Q5575167","display_name":"Goal modeling","level":4,"score":0.4553999900817871},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.45509999990463257},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.4537000060081482},{"id":"https://openalex.org/C2781195723","wikidata":"https://www.wikidata.org/wiki/Q199027","display_name":"Goal setting","level":2,"score":0.3343000113964081},{"id":"https://openalex.org/C2775997480","wikidata":"https://www.wikidata.org/wiki/Q586277","display_name":"Degree (music)","level":2,"score":0.32170000672340393},{"id":"https://openalex.org/C60777511","wikidata":"https://www.wikidata.org/wiki/Q3045002","display_name":"Concept drift","level":3,"score":0.3077000081539154},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.30660000443458557},{"id":"https://openalex.org/C107457646","wikidata":"https://www.wikidata.org/wiki/Q207434","display_name":"Human\u2013computer interaction","level":1,"score":0.2897000014781952},{"id":"https://openalex.org/C195324797","wikidata":"https://www.wikidata.org/wiki/Q33742","display_name":"Natural language","level":2,"score":0.2687000036239624},{"id":"https://openalex.org/C13687954","wikidata":"https://www.wikidata.org/wiki/Q4826847","display_name":"Autonomous agent","level":2,"score":0.25380000472068787}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.1609/aies.v8i1.36541","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aies.v8i1.36541","pdf_url":"https://ojs.aaai.org/index.php/AIES/article/download/36541/38679","source":{"id":"https://openalex.org/S5407048695","display_name":"Proceedings of the AAAI/ACM Conference on AI Ethics and Society","issn_l":"3065-8365","issn":["3065-8365"],"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society","raw_type":"journal-article"},{"id":"pmh:oai:ojs.aaai.org:article/36541","is_oa":false,"landing_page_url":"https://ojs.aaai.org/index.php/AIES/article/view/36541","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"3065-8365","raw_type":"info:eu-repo/semantics/article"}],"best_oa_location":{"id":"doi:10.1609/aies.v8i1.36541","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aies.v8i1.36541","pdf_url":"https://ojs.aaai.org/index.php/AIES/article/download/36541/38679","source":{"id":"https://openalex.org/S5407048695","display_name":"Proceedings of the AAAI/ACM Conference on AI Ethics and Society","issn_l":"3065-8365","issn":["3065-8365"],"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society","raw_type":"journal-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4415230474.pdf","grobid_xml":"https://content.openalex.org/works/W4415230474.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"As":[0],"language":[1],"models":[2,137],"(LMs)":[3],"are":[4,87],"increasingly":[5],"deployed":[6],"as":[7,60,158],"autonomous":[8],"agents,":[9],"their":[10,94],"robust":[11],"adherence":[12,123],"to":[13,48,76,99,155],"human-assigned":[14],"objectives":[15,101],"becomes":[16],"crucial":[17],"for":[18,26,124],"safe":[19],"operation.":[20],"When":[21],"these":[22],"agents":[23,86],"operate":[24],"independently":[25],"extended":[27],"periods":[28],"without":[29],"human":[30],"oversight,":[31],"even":[32],"initially":[33],"well-specified":[34],"goals":[35,61],"may":[36],"gradually":[37],"shift.":[38],"Detecting":[39],"and":[40],"measuring":[41],"goal":[42,78,92,122,142,148],"drift":[43,79,149],"-":[44,56],"an":[45],"agent's":[46],"tendency":[47],"deviate":[49],"from":[50],"its":[51],"original":[52],"objective":[53],"over":[54],"time":[55],"presents":[57],"significant":[58],"challenges,":[59],"can":[62],"shift":[63],"gradually,":[64],"causing":[65],"only":[66],"subtle":[67],"behavioral":[68],"changes.":[69],"This":[70],"paper":[71],"proposes":[72],"a":[73,91],"novel":[74],"approach":[75],"analyzing":[77],"in":[80,129],"LM":[81],"agents.":[82],"In":[83],"our":[84,130],"experiments,":[85],"first":[88],"explicitly":[89],"given":[90],"through":[93,102],"system":[95],"prompt,":[96],"then":[97],"exposed":[98],"competing":[100],"environmental":[103],"pressures.":[104],"We":[105,144],"demonstrate":[106],"that":[107,147],"while":[108],"the":[109,159],"best-performing":[110],"agent":[111],"(a":[112],"scaffolded":[113],"version":[114],"of":[115,141],"Claude":[116],"3.5":[117],"Sonnet)":[118],"maintains":[119],"nearly":[120],"perfect":[121],"more":[125],"than":[126],"100,000":[127],"tokens":[128],"most":[131],"difficult":[132],"evaluation":[133],"setting,":[134],"all":[135],"evaluated":[136],"exhibit":[138],"some":[139],"degree":[140],"drift.":[143],"also":[145],"find":[146],"correlates":[150],"with":[151],"models'":[152],"increasing":[153],"susceptibility":[154],"pattern-matching":[156],"behaviors":[157],"context":[160],"length":[161],"grows.":[162]},"counts_by_year":[{"year":2026,"cited_by_count":5}],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2025-10-16T00:00:00"}