{"id":"https://openalex.org/W7202298173","doi":"https://doi.org/10.48550/arxiv.2608.10171","title":"Generating Attacks for LLMs with GFlowNets","display_name":"Generating Attacks for LLMs with GFlowNets","publication_year":2026,"publication_date":"2026-08-10","ids":{"openalex":"https://openalex.org/W7202298173","doi":"https://doi.org/10.48550/arxiv.2608.10171"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2608.10171","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.10171","pdf_url":"https://arxiv.org/pdf/2608.10171","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":null,"license_id":null,"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/2608.10171","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5080119417","display_name":"Berkay \u00d6z\u00e7am","orcid":"https://orcid.org/0000-0002-6352-1060","observed_orcids":["https://orcid.org/0000-0002-6352-1060"]},"institutions":[{"id":"https://openalex.org/I4210154164","display_name":"Turkcell (Turkey)","ror":"https://ror.org/051vt5y84","country_code":"TR","type":"company","lineage":["https://openalex.org/I4210154164"]}],"countries":["TR"],"is_corresponding":false,"raw_author_name":"Ozcam, Berkay","raw_affiliation_strings":["Cybersecurity R&D Turkcell Istanbul , Turkey"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Cybersecurity R&D Turkcell Istanbul , Turkey","institution_ids":["https://openalex.org/I4210154164"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5146462302","display_name":"\u0130rem \u00d6nen","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I4210154164","display_name":"Turkcell (Turkey)","ror":"https://ror.org/051vt5y84","country_code":"TR","type":"company","lineage":["https://openalex.org/I4210154164"]}],"countries":["TR"],"is_corresponding":false,"raw_author_name":"Onen, Irem","raw_affiliation_strings":["Cybersecurity R&D Turkcell Istanbul , Turkey"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Cybersecurity R&D Turkcell Istanbul , Turkey","institution_ids":["https://openalex.org/I4210154164"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5080453744","display_name":"Mehmet Fatih Amasyal\u0131","orcid":"https://orcid.org/0000-0002-0404-5973","observed_orcids":["https://orcid.org/0000-0002-0404-5973"]},"institutions":[{"id":"https://openalex.org/I4101805","display_name":"Y\u0131ld\u0131z Technical University","ror":"https://ror.org/0547yzj13","country_code":"TR","type":"education","lineage":["https://openalex.org/I4101805"]}],"countries":["TR"],"is_corresponding":false,"raw_author_name":"Amasyali, Mehmet Fatih","raw_affiliation_strings":["Department of Computer Engineering Y\u0131ld\u0131z Technical University Istanbul , Turkey"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Computer Engineering Y\u0131ld\u0131z Technical University Istanbul , Turkey","institution_ids":["https://openalex.org/I4101805"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5087110439","display_name":"Emin \u0130slam Tatl\u0131","orcid":"https://orcid.org/0000-0003-4562-8486","observed_orcids":["https://orcid.org/0000-0003-4562-8486"]},"institutions":[{"id":"https://openalex.org/I4210154164","display_name":"Turkcell (Turkey)","ror":"https://ror.org/051vt5y84","country_code":"TR","type":"company","lineage":["https://openalex.org/I4210154164"]}],"countries":["TR"],"is_corresponding":false,"raw_author_name":"Tatli, Emin Islam","raw_affiliation_strings":["Cybersecurity R&D Turkcell Istanbul , Turkey"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Cybersecurity R&D Turkcell Istanbul , Turkey","institution_ids":["https://openalex.org/I4210154164"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"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":1.0,"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":1.0,"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":0.0,"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/T11424","display_name":"Security and Verification in Computing","score":0.0,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"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":0.8629999756813049},{"id":"https://openalex.org/keywords/adversarial-attacks","display_name":"adversarial attacks","score":0.8709999918937683},{"id":"https://openalex.org/keywords/generative-flow-networks","display_name":"generative flow networks","score":0.9229999780654907},{"id":"https://openalex.org/keywords/red-teaming","display_name":"red teaming","score":0.8790000081062317},{"id":"https://openalex.org/keywords/robustness-analysis","display_name":"robustness analysis","score":0.6869999766349792}],"concepts":[{"id":"https://openalex.org/C37736160","wikidata":"https://www.wikidata.org/wiki/Q1801315","display_name":"Adversarial system","level":2,"score":0.7522000074386597},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.727400004863739},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.6931999921798706},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6604999899864197},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.4399000108242035},{"id":"https://openalex.org/C116834253","wikidata":"https://www.wikidata.org/wiki/Q2039217","display_name":"Identification (biology)","level":2,"score":0.436599999666214},{"id":"https://openalex.org/C19768560","wikidata":"https://www.wikidata.org/wiki/Q320727","display_name":"Dependency (UML)","level":2,"score":0.3977000117301941},{"id":"https://openalex.org/C2781121862","wikidata":"https://www.wikidata.org/wiki/Q256","display_name":"Turkish","level":2,"score":0.33160001039505005},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.31690001487731934},{"id":"https://openalex.org/C112930515","wikidata":"https://www.wikidata.org/wiki/Q4389547","display_name":"Risk analysis (engineering)","level":1,"score":0.30300000309944153},{"id":"https://openalex.org/C140547941","wikidata":"https://www.wikidata.org/wiki/Q7797194","display_name":"Threat model","level":2,"score":0.29840001463890076},{"id":"https://openalex.org/C2780741293","wikidata":"https://www.wikidata.org/wiki/Q4818019","display_name":"Attack patterns","level":3,"score":0.29499998688697815},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.29319998621940613},{"id":"https://openalex.org/C195518309","wikidata":"https://www.wikidata.org/wiki/Q13424265","display_name":"Security testing","level":5,"score":0.28839999437332153},{"id":"https://openalex.org/C128942645","wikidata":"https://www.wikidata.org/wiki/Q1568346","display_name":"Test case","level":3,"score":0.2802000045776367},{"id":"https://openalex.org/C121822524","wikidata":"https://www.wikidata.org/wiki/Q5157582","display_name":"Computer security model","level":2,"score":0.27079999446868896},{"id":"https://openalex.org/C92446256","wikidata":"https://www.wikidata.org/wiki/Q3306762","display_name":"Data validation","level":2,"score":0.2597000002861023},{"id":"https://openalex.org/C80519477","wikidata":"https://www.wikidata.org/wiki/Q3532236","display_name":"Scenario testing","level":3,"score":0.2513999938964844},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.2506999969482422}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2608.10171","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.10171","pdf_url":"https://arxiv.org/pdf/2608.10171","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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2608.10171","doi":"https://doi.org/10.48550/arxiv.2608.10171","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2608.10171","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2608.10171","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.10171","pdf_url":"https://arxiv.org/pdf/2608.10171","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":null,"license_id":null,"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/W7202298173.pdf","grobid_xml":"https://content.openalex.org/works/W7202298173.grobid-xml"},"referenced_works_count":12,"referenced_works":["https://openalex.org/W2969117066","https://openalex.org/W4311991106","https://openalex.org/W4327810158","https://openalex.org/W4385571804","https://openalex.org/W4385894687","https://openalex.org/W4399794780","https://openalex.org/W4399795361","https://openalex.org/W4401306886","https://openalex.org/W4414939355","https://openalex.org/W4415797023","https://openalex.org/W6947991961","https://openalex.org/W7092296780"],"related_works":["https://openalex.org/W4403851788","https://openalex.org/W7157187412","https://openalex.org/W4415313671","https://openalex.org/W4404782310","https://openalex.org/W7117307399","https://openalex.org/W4400434186","https://openalex.org/W4389519367","https://openalex.org/W4416034565","https://openalex.org/W4415317790","https://openalex.org/W4414892798"],"abstract_inverted_index":{"The":[0],"rapid":[1],"advancement":[2],"of":[3,34,174],"Large":[4],"Language":[5],"Models":[6],"(LLMs)":[7],"has":[8,24],"facilitated":[9],"their":[10,91],"ubiquitous":[11],"integration":[12],"into":[13],"various":[14],"domains,":[15],"leading":[16],"to":[17,44,90,110,120,136,150,159,167],"widespread":[18],"adoption.":[19],"However,":[20],"this":[21,98,124],"escalating":[22],"trend":[23],"introduced":[25],"significant":[26],"security":[27,56],"vulnerabilities,":[28],"necessitating":[29],"the":[30,168,179],"identification":[31],"and":[32,58,105,141],"mitigation":[33],"flaws":[35],"arising":[36],"from":[37,86],"malicious":[38],"exploitation.":[39],"Red":[40],"teaming":[41,63,140],"assessments,":[42],"conducted":[43],"evaluate":[45],"model":[46,119,128,135,172],"robustness":[47,145],"through":[48],"diverse":[49],"adversarial":[50,154],"inputs,":[51],"are":[52],"essential":[53],"for":[54],"exposing":[55],"risks":[57],"implementing":[59],"countermeasures.":[60],"Currently,":[61],"red":[62,139],"is":[64,129],"performed":[65],"either":[66],"manually":[67],"by":[68,114],"experts":[69],"or":[70],"automatically":[71],"using":[72],"predefined":[73],"attack":[74,176],"datasets.":[75,96],"Nevertheless,":[76],"manual":[77],"testing":[78],"remains":[79],"time-consuming,":[80],"while":[81],"existing":[82,160],"automated":[83,138],"methods":[84],"suffer":[85],"limited":[87],"creativity":[88],"due":[89],"inherent":[92],"dependency":[93],"on":[94],"fixed":[95],"In":[97],"study,":[99],"we":[100],"propose":[101],"an":[102,126],"automated,":[103],"human-independent,":[104],"adaptive":[106],"approach":[107],"leveraging":[108],"GFlowNets":[109],"identify":[111],"LLM":[112],"vulnerabilities":[113],"utilizing":[115],"one":[116],"large":[117],"language":[118],"test":[121],"another.":[122],"Within":[123],"framework,":[125],"attacker":[127],"trained":[130],"against":[131],"a":[132,143,164,171],"specified":[133],"victim":[134],"perform":[137],"provide":[142],"quantitative":[144],"score.":[146],"This":[147],"research":[148],"aims":[149],"generate":[151],"more":[152],"effective":[153],"attacks":[155],"in":[156,178],"English":[157],"compared":[158],"benchmarks":[161],"and,":[162],"as":[163],"novel":[165],"contribution":[166],"literature,":[169],"introduces":[170],"capable":[173],"generating":[175],"inputs":[177],"Turkish":[180],"language.":[181]},"counts_by_year":[],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2026-08-13T00:00:00"}