{"id":"https://openalex.org/W4414928232","doi":"https://doi.org/10.48550/arxiv.2510.02286","title":"Tree-based Dialogue Reinforced Policy Optimization for Red-Teaming Attacks","display_name":"Tree-based Dialogue Reinforced Policy Optimization for Red-Teaming Attacks","publication_year":2025,"publication_date":"2025-10-02","ids":{"openalex":"https://openalex.org/W4414928232","doi":"https://doi.org/10.48550/arxiv.2510.02286"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2510.02286","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2510.02286","pdf_url":"https://arxiv.org/pdf/2510.02286","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","license_id":"https://openalex.org/licenses/cc-by","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/2510.02286","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":null,"display_name":"Guo, Ruohao","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I130701444","display_name":"Georgia Institute of Technology","ror":"https://ror.org/01zkghx44","country_code":"US","type":"education","lineage":["https://openalex.org/I130701444"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Guo, Ruohao","raw_affiliation_strings":["Georgia Institute of Technology"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Georgia Institute of Technology","institution_ids":["https://openalex.org/I130701444"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5113406990","display_name":"Afshin Oroojlooy","orcid":"https://orcid.org/0000-0001-7829-6145","observed_orcids":["https://orcid.org/0000-0001-7829-6145"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Oroojlooy, Afshin","raw_affiliation_strings":["Oracle AI"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Oracle AI","institution_ids":[]}]},{"author_position":"middle","author":{"id":null,"display_name":"Sridhar, Roshan","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sridhar, Roshan","raw_affiliation_strings":["Oracle AI"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Oracle AI","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5027081783","display_name":"Miguel Ballesteros","orcid":"https://orcid.org/0000-0002-8445-4303","observed_orcids":["https://orcid.org/0000-0002-8445-4303"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ballesteros, Miguel","raw_affiliation_strings":["Oracle AI"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Oracle AI","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5039096905","display_name":"Alan Ritter","orcid":"https://orcid.org/0009-0001-4602-138X","observed_orcids":["https://orcid.org/0009-0001-4602-138X"]},"institutions":[{"id":"https://openalex.org/I130701444","display_name":"Georgia Institute of Technology","ror":"https://ror.org/01zkghx44","country_code":"US","type":"education","lineage":["https://openalex.org/I130701444"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Ritter, Alan","raw_affiliation_strings":["Georgia Institute of Technology"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Georgia Institute of Technology","institution_ids":["https://openalex.org/I130701444"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5023802054","display_name":"Dan Roth","orcid":"https://orcid.org/0009-0002-1447-5173","observed_orcids":["https://orcid.org/0009-0002-1447-5173"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Roth, Dan","raw_affiliation_strings":["Oracle AI"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Oracle AI","institution_ids":[]}]}],"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":6,"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/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"}},{"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"}}],"keywords":[{"id":"https://openalex.org/keywords/large-language-models","display_name":"large language models","score":0.7609999775886536},{"id":"https://openalex.org/keywords/adversarial-attacks","display_name":"adversarial attacks","score":0.7110000252723694},{"id":"https://openalex.org/keywords/tree-search","display_name":"tree search","score":0.9649999737739563},{"id":"https://openalex.org/keywords/ai-safety","display_name":"AI safety","score":0.527999997138977},{"id":"https://openalex.org/keywords/prompt-engineering","display_name":"prompt engineering","score":0.4009999930858612},{"id":"https://openalex.org/keywords/sequential-decision-making","display_name":"sequential decision making","score":0.5600000023841858}],"concepts":[{"id":"https://openalex.org/C2777200299","wikidata":"https://www.wikidata.org/wiki/Q52943","display_name":"Conversation","level":2,"score":0.7135000228881836},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6904000043869019},{"id":"https://openalex.org/C37736160","wikidata":"https://www.wikidata.org/wiki/Q1801315","display_name":"Adversarial system","level":2,"score":0.6722999811172485},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.6571000218391418},{"id":"https://openalex.org/C95713431","wikidata":"https://www.wikidata.org/wiki/Q631425","display_name":"Vulnerability (computing)","level":2,"score":0.6100000143051147},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.49880000948905945},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.41780000925064087},{"id":"https://openalex.org/C2778572836","wikidata":"https://www.wikidata.org/wiki/Q380933","display_name":"Space (punctuation)","level":2,"score":0.40310001373291016},{"id":"https://openalex.org/C113174947","wikidata":"https://www.wikidata.org/wiki/Q2859736","display_name":"Tree (set theory)","level":2,"score":0.37950000166893005},{"id":"https://openalex.org/C2522767166","wikidata":"https://www.wikidata.org/wiki/Q2374463","display_name":"Data science","level":1,"score":0.32600000500679016},{"id":"https://openalex.org/C107457646","wikidata":"https://www.wikidata.org/wiki/Q207434","display_name":"Human\u2013computer interaction","level":1,"score":0.3181000053882599},{"id":"https://openalex.org/C112930515","wikidata":"https://www.wikidata.org/wiki/Q4389547","display_name":"Risk analysis (engineering)","level":1,"score":0.26409998536109924},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.2639999985694885},{"id":"https://openalex.org/C140547941","wikidata":"https://www.wikidata.org/wiki/Q7797194","display_name":"Threat model","level":2,"score":0.25600001215934753},{"id":"https://openalex.org/C136197465","wikidata":"https://www.wikidata.org/wiki/Q1729295","display_name":"Variety (cybernetics)","level":2,"score":0.2540999948978424}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2510.02286","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2510.02286","pdf_url":"https://arxiv.org/pdf/2510.02286","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","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2510.02286","doi":"https://doi.org/10.48550/arxiv.2510.02286","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2510.02286","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","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2510.02286","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2510.02286","pdf_url":"https://arxiv.org/pdf/2510.02286","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","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4414928232.pdf","grobid_xml":"https://content.openalex.org/works/W4414928232.grobid-xml"},"referenced_works_count":26,"referenced_works":["https://openalex.org/W2736601468","https://openalex.org/W2963167310","https://openalex.org/W4327810158","https://openalex.org/W4380353722","https://openalex.org/W4385374425","https://openalex.org/W4385571830","https://openalex.org/W4387687151","https://openalex.org/W4389518965","https://openalex.org/W4391631327","https://openalex.org/W4396822213","https://openalex.org/W4401306886","https://openalex.org/W4401306979","https://openalex.org/W4402671137","https://openalex.org/W4402671495","https://openalex.org/W4405766390","https://openalex.org/W4407690296","https://openalex.org/W4412889939","https://openalex.org/W4412945143","https://openalex.org/W4412945254","https://openalex.org/W4414892798","https://openalex.org/W4415797606","https://openalex.org/W4416035431","https://openalex.org/W4417422257","https://openalex.org/W6948369276","https://openalex.org/W7077869697","https://openalex.org/W7196953391"],"related_works":["https://openalex.org/W7128580605","https://openalex.org/W7128558069","https://openalex.org/W4416036688","https://openalex.org/W6979366114","https://openalex.org/W4414891717","https://openalex.org/W7168029301","https://openalex.org/W6966618268","https://openalex.org/W7166877710","https://openalex.org/W4416156393","https://openalex.org/W6947897211"],"abstract_inverted_index":{"Despite":[0],"recent":[1,103],"rapid":[2],"progress":[3],"in":[4,16],"AI":[5],"safety,":[6],"current":[7],"large":[8],"language":[9],"models":[10,169],"remain":[11],"vulnerable":[12],"to":[13,82,111,115,171],"adversarial":[14],"attacks":[15,113],"multi-turn":[17,79,112,134],"interaction":[18],"settings,":[19],"where":[20],"attackers":[21],"strategically":[22],"adapt":[23],"their":[24],"prompts":[25],"across":[26,166,191],"conversation":[27,95],"turns":[28],"and":[29,58,93],"pose":[30],"a":[31,142],"more":[32,161],"critical":[33,101],"yet":[34],"realistic":[35],"challenge.":[36],"Existing":[37],"approaches":[38],"that":[39,87,105,130,187],"discover":[40],"safety":[41],"vulnerabilities":[42],"either":[43],"rely":[44],"on":[45,65],"manual":[46],"red-teaming":[47],"with":[48,62,127],"human":[49],"experts":[50],"or":[51],"employ":[52],"automated":[53],"methods":[54,70],"using":[55],"pre-defined":[56],"templates":[57],"human-curated":[59],"attack":[60,85,135,180,189],"data,":[61],"most":[63],"focusing":[64],"single-turn":[66,116],"attacks.":[67,117],"However,":[68],"these":[69],"did":[71],"not":[72,158],"explore":[73],"the":[74,139],"vast":[75],"space":[76],"of":[77],"possible":[78],"attacks,":[80],"failing":[81],"consider":[83],"novel":[84],"trajectories":[86],"emerge":[88],"from":[89],"complex":[90],"dialogue":[91,140,185],"dynamics":[92],"strategic":[94],"planning.":[96],"This":[97],"gap":[98],"is":[99],"particularly":[100],"given":[102],"findings":[104],"LLMs":[106],"exhibit":[107],"significantly":[108],"higher":[109,164],"vulnerability":[110],"compared":[114,170],"We":[118],"propose":[119],"DialTree,":[120],"an":[121],"on-policy":[122],"reinforcement":[123],"learning":[124,183],"framework":[125],"integrated":[126],"tree":[128],"search":[129],"autonomously":[131],"discovers":[132],"diverse":[133],"strategies":[136,181],"by":[137,182],"treating":[138],"as":[141],"sequential":[143],"decision-making":[144],"problem,":[145],"enabling":[146],"systematic":[147],"exploration":[148],"without":[149],"manually":[150],"curated":[151],"data.":[152],"Through":[153],"extensive":[154],"experiments,":[155],"our":[156],"approach":[157],"only":[159],"achieves":[160],"than":[162],"44.2%":[163],"ASR":[165],"12":[167],"target":[168],"previous":[172],"state-of-the-art":[173],"approaches,":[174],"but":[175],"also":[176],"effectively":[177],"uncovers":[178],"new":[179],"optimal":[184],"policies":[186],"maximize":[188],"success":[190],"multiple":[192],"turns.":[193]},"counts_by_year":[{"year":2026,"cited_by_count":5},{"year":2025,"cited_by_count":1}],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2025-10-10T00:00:00"}