{"id":"https://openalex.org/W4412886656","doi":"https://doi.org/10.18653/v1/2025.acl-long.238","title":"from Benign import Toxic: Jailbreaking the Language Model via Adversarial Metaphors","display_name":"from Benign import Toxic: Jailbreaking the Language Model via Adversarial Metaphors","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4412886656","doi":"https://doi.org/10.18653/v1/2025.acl-long.238"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.acl-long.238","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.238","pdf_url":"https://aclanthology.org/2025.acl-long.238.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":"Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["arxiv","crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.acl-long.238.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5100411366","display_name":"Yu Yan","orcid":"https://orcid.org/0000-0001-5942-5853"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yu Yan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5032279892","display_name":"Sheng Sun","orcid":"https://orcid.org/0000-0002-2849-0908"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sheng Sun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Zenghao Duan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zenghao Duan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Teli Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Teli Liu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101475946","display_name":"Min Liu","orcid":"https://orcid.org/0000-0002-4952-2577"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Min Liu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5083290750","display_name":"Zhiyi Yin","orcid":"https://orcid.org/0000-0001-8334-3233"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhiyi Yin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5119180880","display_name":"LeiJingyu LeiJingyu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"LeiJingyu LeiJingyu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5100350165","display_name":"Qi Li","orcid":"https://orcid.org/0000-0001-8776-8730"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Qi Li","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":2,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"4785","last_page":"4817"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.7904999852180481,"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":0.7904999852180481,"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/adversarial-system","display_name":"Adversarial system","score":0.8428226113319397},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.6467579007148743},{"id":"https://openalex.org/keywords/natural-language-processing","display_name":"Natural language processing","score":0.4694618284702301},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.42950087785720825}],"concepts":[{"id":"https://openalex.org/C37736160","wikidata":"https://www.wikidata.org/wiki/Q1801315","display_name":"Adversarial system","level":2,"score":0.8428226113319397},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6467579007148743},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.4694618284702301},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.42950087785720825}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.18653/v1/2025.acl-long.238","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.238","pdf_url":"https://aclanthology.org/2025.acl-long.238.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":"Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2503.00038","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2503.00038","pdf_url":"https://arxiv.org/pdf/2503.00038","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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"pmh:doi:10.48550/arxiv.2503.00038","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"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":"Article"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.acl-long.238","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.238","pdf_url":"https://aclanthology.org/2025.acl-long.238.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":"Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4412886656.pdf"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W2502115930","https://openalex.org/W2482350142","https://openalex.org/W4246396837","https://openalex.org/W3126451824","https://openalex.org/W1561927205","https://openalex.org/W3191453585","https://openalex.org/W3204019825"],"abstract_inverted_index":{"Yu":[0],"Yan,":[1],"Sheng":[2],"Sun,":[3],"Zenghao":[4],"Duan,":[5],"Teli":[6],"Liu,":[7,9],"Min":[8],"Zhiyi":[10],"Yin,":[11],"LeiJingyu":[12],"LeiJingyu,":[13],"Qi":[14],"Li.":[15],"Proceedings":[16],"of":[17,22],"the":[18,23],"63rd":[19],"Annual":[20],"Meeting":[21],"Association":[24],"for":[25],"Computational":[26],"Linguistics":[27],"(Volume":[28],"1:":[29],"Long":[30],"Papers).":[31],"2025.":[32]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2025,"cited_by_count":1}],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}