{"id":"https://openalex.org/W7140072463","doi":"https://doi.org/10.18653/v1/2026.findings-eacl.72","title":"Persona Jailbreaking in Large Language Models","display_name":"Persona Jailbreaking in Large Language Models","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7140072463","doi":"https://doi.org/10.18653/v1/2026.findings-eacl.72"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.findings-eacl.72","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-eacl.72","pdf_url":"https://aclanthology.org/2026.findings-eacl.72.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: EACL 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-eacl.72.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5071913667","display_name":"Jivnesh Sandhan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jivnesh Sandhan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5130395653","display_name":"Fei Cheng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Fei Cheng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5130331088","display_name":"Tushar Sandhan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tushar Sandhan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5013357764","display_name":"Yugo Murawaki","orcid":"https://orcid.org/0000-0002-0863-1507"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yugo Murawaki","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":0.19245555,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"1412","last_page":"1430"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T14074","display_name":"Persona Design and Applications","score":0.9175999760627747,"subfield":{"id":"https://openalex.org/subfields/1709","display_name":"Human-Computer Interaction"},"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/T14074","display_name":"Persona Design and Applications","score":0.9175999760627747,"subfield":{"id":"https://openalex.org/subfields/1709","display_name":"Human-Computer Interaction"},"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/T12380","display_name":"Authorship Attribution and Profiling","score":0.006200000178068876,"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/T10028","display_name":"Topic Modeling","score":0.006000000052154064,"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/persona","display_name":"Persona","score":0.8865000009536743},{"id":"https://openalex.org/keywords/focus","display_name":"Focus (optics)","score":0.46869999170303345},{"id":"https://openalex.org/keywords/robustness","display_name":"Robustness (evolution)","score":0.42649999260902405},{"id":"https://openalex.org/keywords/vulnerability","display_name":"Vulnerability (computing)","score":0.4178999960422516},{"id":"https://openalex.org/keywords/narrative","display_name":"Narrative","score":0.39879998564720154},{"id":"https://openalex.org/keywords/metric","display_name":"Metric (unit)","score":0.3880999982357025},{"id":"https://openalex.org/keywords/task","display_name":"Task (project management)","score":0.376800000667572}],"concepts":[{"id":"https://openalex.org/C313442","wikidata":"https://www.wikidata.org/wiki/Q778556","display_name":"Persona","level":2,"score":0.8865000009536743},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6948000192642212},{"id":"https://openalex.org/C192209626","wikidata":"https://www.wikidata.org/wiki/Q190909","display_name":"Focus (optics)","level":2,"score":0.46869999170303345},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.4681999981403351},{"id":"https://openalex.org/C107457646","wikidata":"https://www.wikidata.org/wiki/Q207434","display_name":"Human\u2013computer interaction","level":1,"score":0.44179999828338623},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.42649999260902405},{"id":"https://openalex.org/C95713431","wikidata":"https://www.wikidata.org/wiki/Q631425","display_name":"Vulnerability (computing)","level":2,"score":0.4178999960422516},{"id":"https://openalex.org/C199033989","wikidata":"https://www.wikidata.org/wiki/Q1318295","display_name":"Narrative","level":2,"score":0.39879998564720154},{"id":"https://openalex.org/C176217482","wikidata":"https://www.wikidata.org/wiki/Q860554","display_name":"Metric (unit)","level":2,"score":0.3880999982357025},{"id":"https://openalex.org/C108827166","wikidata":"https://www.wikidata.org/wiki/Q175975","display_name":"Internet privacy","level":1,"score":0.3840000033378601},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.376800000667572},{"id":"https://openalex.org/C153180980","wikidata":"https://www.wikidata.org/wiki/Q19776675","display_name":"Commit","level":2,"score":0.36410000920295715},{"id":"https://openalex.org/C3017944768","wikidata":"https://www.wikidata.org/wiki/Q1450463","display_name":"Poison control","level":2,"score":0.33809998631477356},{"id":"https://openalex.org/C37736160","wikidata":"https://www.wikidata.org/wiki/Q1801315","display_name":"Adversarial system","level":2,"score":0.33469998836517334},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.31540000438690186},{"id":"https://openalex.org/C51929080","wikidata":"https://www.wikidata.org/wiki/Q2425187","display_name":"Codebase","level":3,"score":0.3052999973297119},{"id":"https://openalex.org/C2982912361","wikidata":"https://www.wikidata.org/wiki/Q1851867","display_name":"Mental model","level":2,"score":0.2955999970436096},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.2800999879837036},{"id":"https://openalex.org/C175154964","wikidata":"https://www.wikidata.org/wiki/Q380077","display_name":"Task analysis","level":3,"score":0.25360000133514404},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.2500999867916107}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2026.findings-eacl.72","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-eacl.72","pdf_url":"https://aclanthology.org/2026.findings-eacl.72.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: EACL 2026","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.findings-eacl.72","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-eacl.72","pdf_url":"https://aclanthology.org/2026.findings-eacl.72.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: EACL 2026","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[{"id":"https://openalex.org/F4320320912","display_name":"Ministry of Education, Culture, Sports, Science and Technology","ror":"https://ror.org/048rj2z13"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7140072463.pdf","grobid_xml":"https://content.openalex.org/works/W7140072463.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Large":[0],"Language":[1],"Models":[2],"(LLMs)":[3],"are":[4,22],"increasingly":[5],"deployed":[6],"in":[7,53,87,97,133,140,168,198,207],"domains":[8,144],"such":[9],"as":[10],"education,":[11],"mental":[12,145],"health":[13],"and":[14,19,34,123,136,148,159,200,210],"customer":[15,149],"support,":[16,150],"where":[17],"stable":[18],"consistent":[20],"personas":[21,199],"critical":[23],"for":[24,51,204],"reliability.Yet,":[25],"existing":[26],"studies":[27],"focus":[28],"on":[29],"narrative":[30],"or":[31],"role-playing":[32],"tasks":[33],"overlook":[35],"how":[36],"adversarial":[37],"conversational":[38],"history":[39],"alone":[40],"can":[41],"reshape":[42],"induced":[43],"personas.Black-box":[44],"persona":[45,62,181,206],"manipulation":[46],"remains":[47],"unexplored,":[48],"raising":[49],"concerns":[50],"robustness":[52],"realistic":[54],"interactions.In":[55],"response,":[56],"we":[57,79],"introduce":[58],"the":[59,89,202],"task":[60],"of":[61],"editing,":[63],"which":[64],"adversarially":[65],"steers":[66],"LLM":[67,98],"traits":[68],"through":[69],"user-side":[70],"inputs":[71],"under":[72,191],"a":[73,94,115,165],"black-box,":[74],"inference-only":[75],"setting.To":[76],"this":[77],"end,":[78],"propose":[80],"PHISH":[81,126,151,162],"(Persona":[82],"Hijacking":[83],"via":[84],"Implicit":[85],"Steering":[86],"History),":[88],"first":[90],"framework":[91],"to":[92,108,117],"expose":[93,195],"new":[95,196],"vulnerability":[96],"safety":[99],"that":[100],"embeds":[101],"semantically":[102],"loaded":[103],"cues":[104],"into":[105],"user":[106],"queries":[107],"gradually":[109],"induce":[110],"reverse":[111],"personas.We":[112],"also":[113],"define":[114],"metric":[116],"quantify":[118],"attack":[119],"success.Across":[120],"3":[121],"benchmarks":[122],"8":[124],"LLMs,":[125],"predictably":[127],"shifts":[128],"personas,":[129,154],"triggers":[130],"collateral":[131],"changes":[132],"correlated":[134],"traits,":[135],"exhibits":[137],"stronger":[138],"effects":[139],"multi-turn":[141],"settings.In":[142],"highrisk":[143],"health,":[146],"tutoring,":[147],"reliably":[152],"manipulates":[153],"validated":[155],"by":[156],"both":[157],"human":[158],"LLM-as-Judge":[160],"evaluations.Importantly,":[161],"causes":[163],"only":[164],"small":[166],"reduction":[167],"reasoning":[169],"benchmark":[170],"performance,":[171],"leaving":[172],"overall":[173],"utility":[174],"largely":[175],"intact":[176],"while":[177],"still":[178],"enabling":[179],"significant":[180],"manipulation.While":[182],"current":[183],"guardrails":[184],"offer":[185],"partial":[186],"protection,":[187],"they":[188],"remain":[189],"brittle":[190],"sustained":[192],"attack.Our":[193],"findings":[194],"vulnerabilities":[197],"highlight":[201],"need":[203],"context-resilient":[205],"LLMs.Our":[208],"codebase":[209],"dataset":[211],"is":[212],"available":[213],"at:":[214]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2026-03-24T00:00:00"}