{"id":"https://openalex.org/W7155003121","doi":"https://doi.org/10.48550/arxiv.2604.15789","title":"A Systematic Study of Training-Free Methods for Trustworthy Large Language Models","display_name":"A Systematic Study of Training-Free Methods for Trustworthy Large Language Models","publication_year":2026,"publication_date":"2026-04-17","ids":{"openalex":"https://openalex.org/W7155003121","doi":"https://doi.org/10.48550/arxiv.2604.15789"},"language":null,"primary_location":{"id":"pmh:doi:10.60882/cispa.32771997","is_oa":true,"landing_page_url":"https://arxiv.org/abs/2604.15789","pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"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":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/abs/2604.15789","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5134006658","display_name":"Wai Man Si","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Si, Wai Man","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134084703","display_name":"Mingjie Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Mingjie","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134017661","display_name":"Michael Backes","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Backes, Michael","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5134020161","display_name":"Yang Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Yang","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":null,"has_fulltext":false,"cited_by_count":0,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.506600022315979,"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":0.506600022315979,"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/T12262","display_name":"Hate Speech and Cyberbullying Detection","score":0.08290000259876251,"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/T11636","display_name":"Artificial Intelligence in Healthcare and Education","score":0.06599999964237213,"subfield":{"id":"https://openalex.org/subfields/2718","display_name":"Health Informatics"},"field":{"id":"https://openalex.org/fields/27","display_name":"Medicine"},"domain":{"id":"https://openalex.org/domains/4","display_name":"Health Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/adversarial-system","display_name":"Adversarial system","score":0.708299994468689},{"id":"https://openalex.org/keywords/trustworthiness","display_name":"Trustworthiness","score":0.6618000268936157},{"id":"https://openalex.org/keywords/robustness","display_name":"Robustness (evolution)","score":0.5813999772071838},{"id":"https://openalex.org/keywords/categorization","display_name":"Categorization","score":0.5257999897003174},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.4925999939441681},{"id":"https://openalex.org/keywords/cover","display_name":"Cover (algebra)","score":0.41990000009536743}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7336000204086304},{"id":"https://openalex.org/C37736160","wikidata":"https://www.wikidata.org/wiki/Q1801315","display_name":"Adversarial system","level":2,"score":0.708299994468689},{"id":"https://openalex.org/C153701036","wikidata":"https://www.wikidata.org/wiki/Q659974","display_name":"Trustworthiness","level":2,"score":0.6618000268936157},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.5813999772071838},{"id":"https://openalex.org/C94124525","wikidata":"https://www.wikidata.org/wiki/Q912550","display_name":"Categorization","level":2,"score":0.5257999897003174},{"id":"https://openalex.org/C112930515","wikidata":"https://www.wikidata.org/wiki/Q4389547","display_name":"Risk analysis (engineering)","level":1,"score":0.5213000178337097},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.4925999939441681},{"id":"https://openalex.org/C2522767166","wikidata":"https://www.wikidata.org/wiki/Q2374463","display_name":"Data science","level":1,"score":0.460099995136261},{"id":"https://openalex.org/C2780428219","wikidata":"https://www.wikidata.org/wiki/Q16952335","display_name":"Cover (algebra)","level":2,"score":0.41990000009536743},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.3476000130176544},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.34630000591278076},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3375999927520752},{"id":"https://openalex.org/C539667460","wikidata":"https://www.wikidata.org/wiki/Q2414942","display_name":"Management science","level":1,"score":0.29679998755455017},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.28769999742507935},{"id":"https://openalex.org/C18762648","wikidata":"https://www.wikidata.org/wiki/Q42213","display_name":"Work (physics)","level":2,"score":0.287200003862381},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.2858999967575073},{"id":"https://openalex.org/C140547941","wikidata":"https://www.wikidata.org/wiki/Q7797194","display_name":"Threat model","level":2,"score":0.28290000557899475},{"id":"https://openalex.org/C66024118","wikidata":"https://www.wikidata.org/wiki/Q1122506","display_name":"Computational model","level":2,"score":0.2517000138759613}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:doi:10.60882/cispa.32771997","is_oa":true,"landing_page_url":"https://arxiv.org/abs/2604.15789","pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"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":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},{"id":"doi:10.48550/arxiv.2604.15789","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.15789","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"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","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:doi:10.60882/cispa.32771997","is_oa":true,"landing_page_url":"https://arxiv.org/abs/2604.15789","pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"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":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"As":[0],"Large":[1],"Language":[2],"Models":[3],"(LLMs)":[4],"receive":[5],"increasing":[6],"attention":[7],"and":[8,27,40,72,82,99,112,118,131,156,166,173,183,189,197],"are":[9,61],"being":[10],"deployed":[11],"across":[12,64,162],"various":[13,109,154],"domains,":[14],"their":[15,56,113],"potential":[16],"risks,":[17],"including":[18],"generating":[19],"harmful":[20],"or":[21],"biased":[22],"content,":[23],"producing":[24],"unsupported":[25],"claims,":[26],"exhibiting":[28],"vulnerabilities":[29],"to":[30,51],"adversarial":[31],"attacks,":[32],"have":[33,45],"drawn":[34],"significant":[35],"attention.":[36],"To":[37,85],"enable":[38],"quick":[39],"low-cost":[41],"adaptation,":[42],"training-free":[43,92,106],"methods":[44,60,107,125,158],"recently":[46],"emerged":[47],"as":[48,79],"cost-effective":[49],"alternatives":[50],"post-training":[52],"alignment":[53],"techniques.":[54],"Despite":[55],"promising":[57],"results,":[58],"these":[59,91,124],"evaluated":[62],"inconsistently":[63],"the":[65,88,102,139,186,202],"literature,":[66,188],"cover":[67],"limited":[68],"dimensions":[69],"of":[70,90,104,153],"trustworthiness,":[71,195],"can":[73],"introduce":[74],"undesirable":[75],"side":[76],"effects,":[77],"such":[78],"utility":[80],"degradation":[81],"increased":[83],"brittleness.":[84],"fully":[86],"assess":[87],"impacts":[89],"methods,":[93],"we":[94,148],"take":[95],"a":[96,150],"step":[97],"back":[98],"systematically":[100],"re-evaluate":[101],"effectiveness":[103],"existing":[105,187],"against":[108],"trustworthy":[110],"settings":[111],"influence":[114],"on":[115,134],"utility,":[116,196],"robustness,":[117],"computational":[119],"overhead.":[120],"We":[121,179],"also":[122],"categorize":[123],"into":[126],"three":[127],"levels":[128],"(input,":[129],"internal,":[130],"output)":[132],"based":[133],"where":[135],"they":[136],"intervene":[137],"in":[138,176,185,199],"model's":[140],"information":[141],"flow":[142],"during":[143],"inference.":[144],"Using":[145],"this":[146],"taxonomy,":[147],"conduct":[149],"comprehensive":[151],"analysis":[152,169],"representative":[155],"effective":[157],"from":[159],"each":[160],"level":[161],"different":[163],"LLM":[164],"families":[165],"sizes.":[167],"Our":[168],"highlights":[170],"several":[171],"trade-offs":[172],"unresolved":[174],"challenges":[175],"current":[177],"approaches.":[178],"summarize":[180],"key":[181],"findings":[182],"limitations":[184],"propose":[190],"practical":[191],"recommendations":[192],"for":[193,204],"balancing":[194],"robustness":[198],"LLMs":[200],"without":[201],"need":[203],"additional":[205],"training.":[206]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-04-21T00:00:00"}