{"id":"https://openalex.org/W7164990027","doi":"https://doi.org/10.48550/arxiv.2606.17110","title":"Loss Landscape Poisoning: Targeted Extraction of Unseen Training Data from LLMs","display_name":"Loss Landscape Poisoning: Targeted Extraction of Unseen Training Data from LLMs","publication_year":2026,"publication_date":"2026-06-15","ids":{"openalex":"https://openalex.org/W7164990027","doi":"https://doi.org/10.48550/arxiv.2606.17110"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.17110","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.17110","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":false,"raw_source_name":null,"raw_type":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2606.17110","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5076232590","display_name":"M Al Mamun","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Mamun, Md Abdullah Al","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5038942051","display_name":"Ngoc Phu Doan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Doan, Ngoc Phu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138788797","display_name":"Pedram Zaree","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zaree, Pedram","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138769760","display_name":"Nael Abu-Ghazaleh","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Abu-Ghazaleh, Nael","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":null,"display_name":"Alouani, Ihsen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Alouani, Ihsen","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":0.7770000100135803,"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":0.7770000100135803,"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/T10764","display_name":"Privacy-Preserving Technologies in Data","score":0.09200000017881393,"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":0.02199999988079071,"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/memorization","display_name":"Memorization","score":0.5806000232696533},{"id":"https://openalex.org/keywords/training","display_name":"Training (meteorology)","score":0.5259000062942505},{"id":"https://openalex.org/keywords/leakage","display_name":"Leakage (economics)","score":0.4887999892234802},{"id":"https://openalex.org/keywords/differential-privacy","display_name":"Differential privacy","score":0.4097000062465668},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.40459999442100525},{"id":"https://openalex.org/keywords/training-set","display_name":"Training set","score":0.3930000066757202},{"id":"https://openalex.org/keywords/information-leakage","display_name":"Information leakage","score":0.3889999985694885},{"id":"https://openalex.org/keywords/data-loss","display_name":"Data loss","score":0.3702000081539154}],"concepts":[{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.6265000104904175},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5982000231742859},{"id":"https://openalex.org/C30038468","wikidata":"https://www.wikidata.org/wiki/Q4354775","display_name":"Memorization","level":2,"score":0.5806000232696533},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.5259000062942505},{"id":"https://openalex.org/C2777042071","wikidata":"https://www.wikidata.org/wiki/Q6509304","display_name":"Leakage (economics)","level":2,"score":0.4887999892234802},{"id":"https://openalex.org/C23130292","wikidata":"https://www.wikidata.org/wiki/Q5275358","display_name":"Differential privacy","level":2,"score":0.4097000062465668},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.40459999442100525},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.3930000066757202},{"id":"https://openalex.org/C2779201187","wikidata":"https://www.wikidata.org/wiki/Q2775060","display_name":"Information leakage","level":2,"score":0.3889999985694885},{"id":"https://openalex.org/C108827166","wikidata":"https://www.wikidata.org/wiki/Q175975","display_name":"Internet privacy","level":1,"score":0.3813000023365021},{"id":"https://openalex.org/C193519340","wikidata":"https://www.wikidata.org/wiki/Q891179","display_name":"Data loss","level":2,"score":0.3702000081539154},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.3499999940395355},{"id":"https://openalex.org/C99221444","wikidata":"https://www.wikidata.org/wiki/Q1532069","display_name":"Private information retrieval","level":2,"score":0.3296000063419342},{"id":"https://openalex.org/C165609540","wikidata":"https://www.wikidata.org/wiki/Q1172486","display_name":"Data breach","level":2,"score":0.3294000029563904},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.320499986410141},{"id":"https://openalex.org/C2992525071","wikidata":"https://www.wikidata.org/wiki/Q50818671","display_name":"Federated learning","level":2,"score":0.31380000710487366},{"id":"https://openalex.org/C2777466982","wikidata":"https://www.wikidata.org/wiki/Q5227287","display_name":"Data extraction","level":3,"score":0.2883000075817108},{"id":"https://openalex.org/C67186912","wikidata":"https://www.wikidata.org/wiki/Q367664","display_name":"Data modeling","level":2,"score":0.28760001063346863},{"id":"https://openalex.org/C2988416141","wikidata":"https://www.wikidata.org/wiki/Q6031139","display_name":"Information loss","level":2,"score":0.27970001101493835},{"id":"https://openalex.org/C123201435","wikidata":"https://www.wikidata.org/wiki/Q456632","display_name":"Information privacy","level":2,"score":0.27869999408721924},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.27379998564720154},{"id":"https://openalex.org/C137822555","wikidata":"https://www.wikidata.org/wiki/Q2587068","display_name":"Information sensitivity","level":2,"score":0.267300009727478},{"id":"https://openalex.org/C47487241","wikidata":"https://www.wikidata.org/wiki/Q5227230","display_name":"Data access","level":2,"score":0.26570001244544983}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.17110","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.17110","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":"doi:10.48550/arxiv.2606.17110","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.17110","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":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[{"score":0.5035994648933411,"display_name":"Peace, Justice and strong institutions","id":"https://metadata.un.org/sdg/16"}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Large":[0],"Language":[1],"Models":[2],"are":[3],"increasingly":[4],"trained":[5,179],"on":[6,114],"proprietary":[7],"or":[8],"sensitive":[9],"data,":[10],"from":[11],"private":[12],"healthcare":[13],"and":[14,72,138,142,161],"financial":[15],"records":[16],"to":[17,101,120,157,180],"user":[18],"conversations":[19],"containing":[20],"secrets.":[21],"Ensuring":[22],"the":[23,50,55,70,86,92,108,118,122,125,149,171,176,193],"privacy":[24,152,199],"of":[25,49,57],"such":[26,75],"data":[27,52],"against":[28],"extraction":[29],"attacks":[30],"has":[31],"become":[32],"a":[33,47,58,81,103,187],"central":[34],"concern.":[35],"In":[36],"this":[37],"paper,":[38],"we":[39,185],"ask":[40],"whether":[41],"an":[42],"attacker":[43],"who":[44],"can":[45,53,77],"poison":[46],"portion":[48],"training":[51],"facilitate":[54],"leakage":[56,76,153],"separate":[59],"target":[60,93,123],"record":[61],"they":[62],"have":[63],"no":[64,135],"access":[65],"to.":[66],"We":[67,146,168],"answer":[68],"in":[69,129],"affirmative":[71],"show":[73,169],"that":[74,84,99,148,170,190],"be":[78,181],"induced":[79],"by":[80,111],"poisoning":[82,100],"mechanism":[83],"reshapes":[85],"model's":[87],"local":[88],"loss":[89,105,113,194],"landscape":[90,195],"around":[91],"completion.":[94],"Our":[95],"key":[96],"insight":[97],"is":[98,173,178],"create":[102],"sharp":[104],"minimum":[106],"at":[107],"target,":[109],"surrounded":[110],"elevated":[112],"nearby":[115],"alternatives,":[116],"forces":[117],"model":[119,177],"memorize":[121],"as":[124],"unique":[126],"low-loss":[127],"solution":[128],"its":[130],"neighborhood.":[131],"The":[132],"attack":[133,150,172,189],"requires":[134],"architectural":[136],"changes,":[137],"generalizes":[139],"across":[140,154],"centralized":[141],"federated":[143],"learning":[144],"settings.":[145],"demonstrate":[147],"amplifies":[151],"language":[155],"(up":[156,164],"100%":[158],"successful":[159,166],"extraction),":[160],"vision-language":[162],"models":[163],"90%":[165],"extraction).":[167],"thwarted":[174],"when":[175],"differentially":[182],"private.":[183],"However,":[184],"introduce":[186],"new":[188],"directly":[191],"probes":[192],"bypassing":[196],"even":[197],"differential":[198],"defenses.":[200]},"counts_by_year":[],"updated_date":"2026-07-21T08:15:58.654021","created_date":"2026-06-18T00:00:00"}