{"id":"https://openalex.org/W4412888727","doi":"https://doi.org/10.18653/v1/2025.findings-acl.257","title":"Better Red Teaming via Searching with Large Language Model","display_name":"Better Red Teaming via Searching with Large Language Model","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4412888727","doi":"https://doi.org/10.18653/v1/2025.findings-acl.257"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.findings-acl.257","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.257","pdf_url":"https://aclanthology.org/2025.findings-acl.257.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: ACL 2025","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.findings-acl.257.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5101794048","display_name":"Yongkang Chen","orcid":"https://orcid.org/0000-0002-8070-9858"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yongkang Chen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5057733392","display_name":"Chongyang Zhao","orcid":"https://orcid.org/0000-0002-2942-1218"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chongyang Zhao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5119181471","display_name":"Jianwentian Jianwentian","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jianwentian Jianwentian","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Guiling Cao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Guiling Cao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100444602","display_name":"Hu Li","orcid":"https://orcid.org/0000-0002-0420-855X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hu Li","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5053431750","display_name":"Xiaohui Kuang","orcid":"https://orcid.org/0000-0003-3816-402X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xiaohui Kuang","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":1.673,"has_fulltext":true,"cited_by_count":1,"citation_normalized_percentile":{"value":0.83433545,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":94,"max":97},"biblio":{"volume":null,"issue":null,"first_page":"4968","last_page":"4984"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.7235000133514404,"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/T10028","display_name":"Topic Modeling","score":0.7235000133514404,"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/computer-science","display_name":"Computer science","score":0.7402020692825317},{"id":"https://openalex.org/keywords/programming-language","display_name":"Programming language","score":0.3638905882835388},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.3209882080554962}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7402020692825317},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.3638905882835388},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3209882080554962}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.findings-acl.257","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.257","pdf_url":"https://aclanthology.org/2025.findings-acl.257.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: ACL 2025","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.findings-acl.257","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.257","pdf_url":"https://aclanthology.org/2025.findings-acl.257.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: ACL 2025","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4412888727.pdf","grobid_xml":"https://content.openalex.org/works/W4412888727.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W2390279801","https://openalex.org/W4391913857","https://openalex.org/W2358668433","https://openalex.org/W4396701345","https://openalex.org/W2376932109","https://openalex.org/W2001405890","https://openalex.org/W4396696052"],"abstract_inverted_index":{"The":[0],"safe":[1],"deployment":[2],"of":[3,27,60,66,74,79,114],"large":[4],"language":[5],"models":[6],"(LLMs)":[7],"necessitates":[8],"comprehensive":[9],"safety":[10],"evaluations":[11],"through":[12],"red":[13],"teaming.However,":[14],"existing":[15],"methods":[16],"face":[17],"challenges":[18],"in":[19],"managing":[20],"semantic":[21],"intricacies":[22],"and":[23,77,94],"optimizing":[24,87],"the":[25,28,64,72,83,98],"efficiency":[26],"search":[29,84],"process.To":[30],"overcome":[31],"these":[32],"limitations,":[33],"we":[34,96],"propose":[35],"Better":[36],"Red":[37],"Teaming":[38],"(BRT)-an":[39],"innovative":[40],"framework":[41],"that":[42,118],"reconceptualizes":[43],"test":[44],"case":[45],"generation":[46],"as":[47,68],"a":[48,112],"strategic":[49],"planning":[50],"problem,":[51],"leveraging":[52],"Monte":[53],"Carlo":[54],"Tree":[55],"Search":[56],"(MCTS).A":[57],"notable":[58],"advancement":[59],"our":[61],"approach":[62],"is":[63],"incorporation":[65],"LLMs":[67],"world":[69,99],"models,":[70],"enabling":[71],"prediction":[73],"state":[75],"transitions":[76],"simulation":[78],"long-term":[80],"outcomes":[81],"throughout":[82],"process.By":[85],"jointly":[86],"objectives":[88],"related":[89],"to":[90,102],"conditional":[91],"mutual":[92],"information":[93],"diversity,":[95],"improve":[97],"model's":[100],"capacity":[101],"follow":[103],"actions":[104],"while":[105],"maintaining":[106],"output":[107],"diversity.Extensive":[108],"experiments":[109],"conducted":[110],"across":[111],"range":[113],"LLM":[115],"architectures":[116],"demonstrate":[117],"BRT":[119],"achieves":[120],"state-of-the-art":[121],"attack":[122],"success":[123],"rates":[124],"without":[125],"sacrificing":[126],"computational":[127],"efficiency.":[128]},"counts_by_year":[{"year":2026,"cited_by_count":1}],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2025-10-10T00:00:00"}