{"id":"https://openalex.org/W7204198583","doi":"https://doi.org/10.48550/arxiv.2608.23264","title":"Hidden in the Request: Explaining Unethical LLM Compliance through Token Relevance","display_name":"Hidden in the Request: Explaining Unethical LLM Compliance through Token Relevance","publication_year":2026,"publication_date":"2026-08-24","ids":{"openalex":"https://openalex.org/W7204198583","doi":"https://doi.org/10.48550/arxiv.2608.23264"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2608.23264","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.23264","pdf_url":"https://arxiv.org/pdf/2608.23264","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by-nc-sa","license_id":"https://openalex.org/licenses/cc-by-nc-sa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2608.23264","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5148152055","display_name":"Or Biton","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Biton, Or","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5148084028","display_name":"Tomer Krichli","orcid":"https://orcid.org/0009-0005-5461-9030","observed_orcids":["https://orcid.org/0009-0005-5461-9030"]},"institutions":[{"id":"https://openalex.org/I174306211","display_name":"Technion \u2013 Israel Institute of Technology","ror":"https://ror.org/03qryx823","country_code":"IL","type":"education","lineage":["https://openalex.org/I174306211"]}],"countries":["IL"],"is_corresponding":false,"raw_author_name":"Krichli, Tomer","raw_affiliation_strings":["Faculty of Electrical and Computer Engineering , Technion , Haifa , Israel"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Faculty of Electrical and Computer Engineering , Technion , Haifa , Israel","institution_ids":["https://openalex.org/I174306211"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5120450007","display_name":"Itai Allouche","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I174306211","display_name":"Technion \u2013 Israel Institute of Technology","ror":"https://ror.org/03qryx823","country_code":"IL","type":"education","lineage":["https://openalex.org/I174306211"]}],"countries":["IL"],"is_corresponding":false,"raw_author_name":"Allouche, Itai","raw_affiliation_strings":["Faculty of Electrical and Computer Engineering , Technion , Haifa , Israel"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Faculty of Electrical and Computer Engineering , Technion , Haifa , Israel","institution_ids":["https://openalex.org/I174306211"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5008847407","display_name":"Joseph Keshet","orcid":"https://orcid.org/0000-0003-2332-5783","observed_orcids":["https://orcid.org/0000-0003-2332-5783"]},"institutions":[{"id":"https://openalex.org/I174306211","display_name":"Technion \u2013 Israel Institute of Technology","ror":"https://ror.org/03qryx823","country_code":"IL","type":"education","lineage":["https://openalex.org/I174306211"]}],"countries":["IL"],"is_corresponding":false,"raw_author_name":"Keshet, Joseph","raw_affiliation_strings":["Faculty of Electrical and Computer Engineering , Technion , Haifa , Israel"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Faculty of Electrical and Computer Engineering , Technion , Haifa , Israel","institution_ids":["https://openalex.org/I174306211"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":[],"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":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10883","display_name":"Ethics and Social Impacts of AI","score":0.7153000235557556,"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/T10883","display_name":"Ethics and Social Impacts of AI","score":0.7153000235557556,"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":0.19020000100135803,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.06520000100135803,"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/large-language-models","display_name":"large language models","score":0.8190000057220459},{"id":"https://openalex.org/keywords/unethical-behavior","display_name":"unethical behavior","score":0.48100000619888306},{"id":"https://openalex.org/keywords/alignment-failure","display_name":"alignment failure","score":0.6290000081062317},{"id":"https://openalex.org/keywords/layer-wise-relevance-propagation","display_name":"layer-wise relevance propagation","score":0.7929999828338623},{"id":"https://openalex.org/keywords/attribution-bias","display_name":"attribution bias","score":0.8360000252723694},{"id":"https://openalex.org/keywords/safety-interventions","display_name":"safety interventions","score":0.6110000014305115}],"concepts":[{"id":"https://openalex.org/C158154518","wikidata":"https://www.wikidata.org/wiki/Q7310970","display_name":"Relevance (law)","level":2,"score":0.7544000148773193},{"id":"https://openalex.org/C2781265381","wikidata":"https://www.wikidata.org/wiki/Q5710255","display_name":"Helpfulness","level":2,"score":0.6366000175476074},{"id":"https://openalex.org/C2781460075","wikidata":"https://www.wikidata.org/wiki/Q1399332","display_name":"Compliance (psychology)","level":2,"score":0.5842999815940857},{"id":"https://openalex.org/C15744967","wikidata":"https://www.wikidata.org/wiki/Q9418","display_name":"Psychology","level":0,"score":0.5551000237464905},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.49079999327659607},{"id":"https://openalex.org/C77805123","wikidata":"https://www.wikidata.org/wiki/Q161272","display_name":"Social psychology","level":1,"score":0.48240000009536743},{"id":"https://openalex.org/C2777267654","wikidata":"https://www.wikidata.org/wiki/Q3519023","display_name":"Test (biology)","level":2,"score":0.4507000148296356},{"id":"https://openalex.org/C27415008","wikidata":"https://www.wikidata.org/wiki/Q7256382","display_name":"Psychological intervention","level":2,"score":0.4320000112056732},{"id":"https://openalex.org/C2776654903","wikidata":"https://www.wikidata.org/wiki/Q2601463","display_name":"SAFER","level":2,"score":0.43160000443458557},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.4097000062465668},{"id":"https://openalex.org/C180747234","wikidata":"https://www.wikidata.org/wiki/Q23373","display_name":"Cognitive psychology","level":1,"score":0.40630000829696655},{"id":"https://openalex.org/C75291252","wikidata":"https://www.wikidata.org/wiki/Q1315756","display_name":"TRACE (psycholinguistics)","level":2,"score":0.39899998903274536},{"id":"https://openalex.org/C2780980858","wikidata":"https://www.wikidata.org/wiki/Q110022","display_name":"Dual (grammatical number)","level":2,"score":0.3813000023365021},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.37450000643730164},{"id":"https://openalex.org/C2777536939","wikidata":"https://www.wikidata.org/wiki/Q25054194","display_name":"Goal pursuit","level":2,"score":0.351500004529953},{"id":"https://openalex.org/C143299363","wikidata":"https://www.wikidata.org/wiki/Q900584","display_name":"Attribution","level":2,"score":0.34220001101493835},{"id":"https://openalex.org/C120936955","wikidata":"https://www.wikidata.org/wiki/Q2155640","display_name":"Empirical research","level":2,"score":0.3402999937534332},{"id":"https://openalex.org/C12713177","wikidata":"https://www.wikidata.org/wiki/Q1900281","display_name":"Perspective (graphical)","level":2,"score":0.3377000093460083},{"id":"https://openalex.org/C2780695315","wikidata":"https://www.wikidata.org/wiki/Q3799040","display_name":"Unobservable","level":2,"score":0.31619998812675476},{"id":"https://openalex.org/C2781310500","wikidata":"https://www.wikidata.org/wiki/Q1231428","display_name":"Persuasion","level":2,"score":0.31290000677108765},{"id":"https://openalex.org/C108827166","wikidata":"https://www.wikidata.org/wiki/Q175975","display_name":"Internet privacy","level":1,"score":0.31119999289512634},{"id":"https://openalex.org/C75630572","wikidata":"https://www.wikidata.org/wiki/Q538904","display_name":"Applied psychology","level":1,"score":0.3059000074863434},{"id":"https://openalex.org/C2779267917","wikidata":"https://www.wikidata.org/wiki/Q170028","display_name":"Deception","level":2,"score":0.2994000017642975},{"id":"https://openalex.org/C166052673","wikidata":"https://www.wikidata.org/wiki/Q83021","display_name":"Empirical evidence","level":2,"score":0.29910001158714294},{"id":"https://openalex.org/C112930515","wikidata":"https://www.wikidata.org/wiki/Q4389547","display_name":"Risk analysis (engineering)","level":1,"score":0.29409998655319214},{"id":"https://openalex.org/C61797465","wikidata":"https://www.wikidata.org/wiki/Q1188986","display_name":"Term (time)","level":2,"score":0.289900004863739},{"id":"https://openalex.org/C158600405","wikidata":"https://www.wikidata.org/wiki/Q5054566","display_name":"Causal inference","level":2,"score":0.2897999882698059},{"id":"https://openalex.org/C18762648","wikidata":"https://www.wikidata.org/wiki/Q42213","display_name":"Work (physics)","level":2,"score":0.2892000079154968},{"id":"https://openalex.org/C11671645","wikidata":"https://www.wikidata.org/wiki/Q5054567","display_name":"Causal model","level":2,"score":0.2883000075817108},{"id":"https://openalex.org/C2779231336","wikidata":"https://www.wikidata.org/wiki/Q7534724","display_name":"Sketch","level":2,"score":0.2838999927043915},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.2800999879837036},{"id":"https://openalex.org/C2776325391","wikidata":"https://www.wikidata.org/wiki/Q6917865","display_name":"Motivated reasoning","level":3,"score":0.2770000100135803},{"id":"https://openalex.org/C339426","wikidata":"https://www.wikidata.org/wiki/Q1151839","display_name":"Prospect theory","level":2,"score":0.26649999618530273},{"id":"https://openalex.org/C64357122","wikidata":"https://www.wikidata.org/wiki/Q1149766","display_name":"Causality (physics)","level":2,"score":0.26339998841285706},{"id":"https://openalex.org/C2780385302","wikidata":"https://www.wikidata.org/wiki/Q367158","display_name":"Protocol (science)","level":3,"score":0.2603999972343445},{"id":"https://openalex.org/C109359841","wikidata":"https://www.wikidata.org/wiki/Q728944","display_name":"Inclusion (mineral)","level":2,"score":0.2587999999523163},{"id":"https://openalex.org/C79897977","wikidata":"https://www.wikidata.org/wiki/Q5054568","display_name":"Causal chain","level":2,"score":0.25270000100135803}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2608.23264","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.23264","pdf_url":"https://arxiv.org/pdf/2608.23264","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by-nc-sa","license_id":"https://openalex.org/licenses/cc-by-nc-sa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2608.23264","doi":"https://doi.org/10.48550/arxiv.2608.23264","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2608.23264","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by-nc-sa","license_id":"https://openalex.org/licenses/cc-by-nc-sa","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2608.23264","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.23264","pdf_url":"https://arxiv.org/pdf/2608.23264","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by-nc-sa","license_id":"https://openalex.org/licenses/cc-by-nc-sa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7204198583.pdf","grobid_xml":"https://content.openalex.org/works/W7204198583.grobid-xml"},"referenced_works_count":8,"referenced_works":["https://openalex.org/W3215698779","https://openalex.org/W4223908421","https://openalex.org/W4390041933","https://openalex.org/W4393905739","https://openalex.org/W4403796339","https://openalex.org/W4412505606","https://openalex.org/W7124165156","https://openalex.org/W7160245013"],"related_works":["https://openalex.org/W4392182312","https://openalex.org/W7164809947","https://openalex.org/W4389599346","https://openalex.org/W4414590635","https://openalex.org/W7166559911","https://openalex.org/W7117322593","https://openalex.org/W7128541643","https://openalex.org/W4408303513","https://openalex.org/W4415981234","https://openalex.org/W7166861570"],"abstract_inverted_index":{"Although":[0],"Large":[1],"Language":[2],"Models":[3],"(LLMs)":[4],"are":[5],"aligned":[6],"to":[7,21,32,53,90,130,149],"optimize":[8],"for":[9,69],"both":[10],"helpfulness":[11],"and":[12,66],"harmlessness,":[13],"these":[14,42,156],"dual":[15],"objectives":[16],"may":[17],"conflict,":[18],"inevitably":[19],"leading":[20],"alignment":[22],"failures.":[23,167],"This":[24],"work":[25],"systematically":[26],"investigates":[27],"instances":[28],"where":[29],"LLMs":[30,54],"fail":[31],"exhibit":[33],"ethical":[34],"behavior.":[35],"To":[36,133],"understand":[37],"the":[38,78,94,112],"underlying":[39,113],"mechanics":[40],"of":[41],"vulnerabilities,":[43],"we":[44,86,121,136],"introduce":[45,137],"a":[46],"probing":[47],"methodology":[48],"that":[49,73,126,142,155],"presents":[50],"unethical":[51,114],"scenarios":[52],"in":[55,77,165],"three":[56],"distinct":[57],"structural":[58],"modalities:":[59],"objective":[60],"classification":[61],"tasks,":[62],"subjective":[63],"first-person":[64],"statements,":[65],"direct":[67],"requests":[68],"assistance.":[70],"We":[71,124],"find":[72],"model":[74,95],"performance":[75],"degrades":[76],"request-for-assistance-based":[79],"form.":[80],"Using":[81],"Layer-wise":[82],"Relevance":[83],"Propagation":[84],"(LRP),":[85],"trace":[87],"this":[88,127],"discrepancy":[89],"an":[91],"attribution":[92],"bias:":[93],"places":[96],"greater":[97],"emphasis":[98],"on":[99,109],"benign":[100],"task-framing":[101],"tokens":[102,110],"(e.g.,":[103,116],"\"Can":[104],"you":[105],"help":[106],"me...\")":[107],"than":[108],"signaling":[111],"behavior":[115],"\"without":[117],"getting":[118],"caught\"),":[119],"which":[120],"term":[122],"cue-tokens.":[123],"hypothesize":[125],"under-attribution":[128],"contributes":[129],"harmful":[131],"compliance.":[132],"test":[134],"this,":[135],"two":[138],"LRP-guided":[139],"decoding":[140],"methods":[141],"steer":[143],"generation":[144],"toward":[145],"trajectories":[146],"more":[147],"relevant":[148],"cue":[150],"tokens.":[151],"Empirical":[152],"evaluations":[153],"show":[154],"interventions":[157],"promote":[158],"safer":[159],"responses,":[160],"supporting":[161],"cue-token":[162],"attribution's":[163],"role":[164],"compliance":[166]},"counts_by_year":[{"year":2026,"cited_by_count":1}],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2026-08-26T00:00:00"}
