{"id":"https://openalex.org/W7166802434","doi":"https://doi.org/10.18653/v1/2026.findings-acl.1797","title":"Measuring Watermarking under Jailbreaking: ASR Inflation and Goal-Compliance Mismatch","display_name":"Measuring Watermarking under Jailbreaking: ASR Inflation and Goal-Compliance Mismatch","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7166802434","doi":"https://doi.org/10.18653/v1/2026.findings-acl.1797"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2026.findings-acl.1797","doi":"https://doi.org/10.18653/v1/2026.findings-acl.1797","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1797","pdf_url":"https://aclanthology.org/2026.findings-acl.1797.pdf","source":{"id":"https://openalex.org/S4363605144","display_name":"Findings of the Association for Computational Linguistics: ACL (Findings of ACL)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"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 2026","raw_type":"proceedings-article"},"volume":null,"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"hybrid","oa_url":"https://aclanthology.org/2026.findings-acl.1797.pdf","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5139760153","display_name":"Sungwoo Han","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I196345858","display_name":"Chungnam National University","ror":"https://ror.org/0227as991","country_code":"KR","type":"education","lineage":["https://openalex.org/I196345858"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Sungwoo Han","raw_affiliation_strings":["Chungnam National University ,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Chungnam National University ,","institution_ids":["https://openalex.org/I196345858"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139721635","display_name":"Sangjun Moon","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I196345858","display_name":"Chungnam National University","ror":"https://ror.org/0227as991","country_code":"KR","type":"education","lineage":["https://openalex.org/I196345858"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Sangjun Moon","raw_affiliation_strings":["Chungnam National University ,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Chungnam National University ,","institution_ids":["https://openalex.org/I196345858"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5028543952","display_name":"Jingun Kwon","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I196345858","display_name":"Chungnam National University","ror":"https://ror.org/0227as991","country_code":"KR","type":"education","lineage":["https://openalex.org/I196345858"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Jingun Kwon","raw_affiliation_strings":["Chungnam National University ,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Chungnam National University ,","institution_ids":["https://openalex.org/I196345858"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5141039303","display_name":"Hidetaka Kamigaito","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I75917431","display_name":"Nara Institute of Science and Technology","ror":"https://ror.org/05bhada84","country_code":"JP","type":"education","lineage":["https://openalex.org/I75917431"]}],"countries":["JP"],"is_corresponding":false,"raw_author_name":"Hidetaka Kamigaito","raw_affiliation_strings":["Nara Institute of Science and Technology (NAIST)"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Nara Institute of Science and Technology (NAIST)","institution_ids":["https://openalex.org/I75917431"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5035876897","display_name":"Manabu Okumura","orcid":"https://orcid.org/0009-0001-7730-1536","observed_orcids":["https://orcid.org/0009-0001-7730-1536"]},"institutions":[{"id":"https://openalex.org/I4400009020","display_name":"Institute of Science Tokyo","ror":"https://ror.org/05dqf9946","country_code":"JP","type":"education","lineage":["https://openalex.org/I4400009020"]}],"countries":["JP"],"is_corresponding":false,"raw_author_name":"Manabu Okumura","raw_affiliation_strings":["Institute of Science Tokyo"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Institute of Science Tokyo","institution_ids":["https://openalex.org/I4400009020"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":3,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.43732209,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"36071","last_page":"36083"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.9643999934196472,"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":0.9643999934196472,"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/T10388","display_name":"Advanced Steganography and Watermarking Techniques","score":0.02319999970495701,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/T12122","display_name":"Physical Unclonable Functions (PUFs) and Hardware Security","score":0.002099999925121665,"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/watermarking","display_name":"watermarking","score":0.7889999747276306},{"id":"https://openalex.org/keywords/jailbreaking","display_name":"jailbreaking","score":0.8880000114440918},{"id":"https://openalex.org/keywords/large-language-models","display_name":"large language models","score":0.9150000214576721},{"id":"https://openalex.org/keywords/source-attribution","display_name":"source attribution","score":0.8610000014305115},{"id":"https://openalex.org/keywords/attack-success-rate","display_name":"attack success rate","score":0.699999988079071},{"id":"https://openalex.org/keywords/model-alignment","display_name":"model alignment","score":0.4259999990463257},{"id":"https://openalex.org/keywords/human-evaluation","display_name":"human evaluation","score":0.4620000123977661}],"concepts":[{"id":"https://openalex.org/C150817343","wikidata":"https://www.wikidata.org/wiki/Q875932","display_name":"Digital watermarking","level":3,"score":0.6789000034332275},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.42500001192092896},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3790000081062317},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.36480000615119934},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.3521000146865845},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.34860000014305115},{"id":"https://openalex.org/C200941418","wikidata":"https://www.wikidata.org/wiki/Q273508","display_name":"Inflation (cosmology)","level":2,"score":0.3249000012874603},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.28439998626708984},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.27549999952316284},{"id":"https://openalex.org/C2780009758","wikidata":"https://www.wikidata.org/wiki/Q6804172","display_name":"Measure (data warehouse)","level":2,"score":0.26930001378059387},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.2624000012874603},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.257999986410141},{"id":"https://openalex.org/C78548338","wikidata":"https://www.wikidata.org/wiki/Q2493","display_name":"Data compression","level":2,"score":0.2502000033855438}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2026.findings-acl.1797","doi":"https://doi.org/10.18653/v1/2026.findings-acl.1797","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1797","pdf_url":"https://aclanthology.org/2026.findings-acl.1797.pdf","source":{"id":"https://openalex.org/S4363605144","display_name":"Findings of the Association for Computational Linguistics: ACL (Findings of ACL)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"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 2026","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.findings-acl.1797","doi":"https://doi.org/10.18653/v1/2026.findings-acl.1797","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1797","pdf_url":"https://aclanthology.org/2026.findings-acl.1797.pdf","source":{"id":"https://openalex.org/S4363605144","display_name":"Findings of the Association for Computational Linguistics: ACL (Findings of ACL)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"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 2026","raw_type":"proceedings-article"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[{"id":"https://openalex.org/F4320320671","display_name":"National Research Foundation","ror":"https://ror.org/05s0g1g46"},{"id":"https://openalex.org/F4320322120","display_name":"National Research Foundation of Korea","ror":"https://ror.org/013aysd81"},{"id":"https://openalex.org/F4320328359","display_name":"Ministry of Science and ICT, South Korea","ror":"https://ror.org/01wpjm123"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7166802434.pdf","grobid_xml":"https://content.openalex.org/works/W7166802434.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4415800537","https://openalex.org/W6997445597","https://openalex.org/W4416033951","https://openalex.org/W4402672097","https://openalex.org/W4401043209","https://openalex.org/W7140112807","https://openalex.org/W4416035049","https://openalex.org/W7165959732","https://openalex.org/W4388555465","https://openalex.org/W4392971839"],"abstract_inverted_index":{"Recently,":[0],"watermarking":[1,20,48,62,80],"has":[2],"attracted":[3],"growing":[4],"attention":[5],"as":[6],"a":[7],"practical":[8],"technique":[9],"for":[10],"source":[11],"attribution":[12],"of":[13],"machine-generated":[14],"text.However,":[15],"most":[16],"prior":[17],"work":[18],"studies":[19],"under":[21,27,54,89],"benign":[22],"prompts,":[23],"while":[24],"its":[25],"behavior":[26],"jailbreaking":[28,35],"prompts":[29],"remains":[30],"underexplored.This":[31],"gap":[32],"matters":[33],"because":[34],"can":[36,81,127],"bypass":[37],"safety":[38],"policies":[39],"and":[40,71,76,133,144],"shift":[41],"the":[42,92,102],"generation":[43],"regime,":[44],"raising":[45],"concerns":[46],"that":[47,79,99,124],"may":[49,134],"interact":[50],"with":[51,91],"model":[52],"alignment":[53],"attack.To":[55],"address":[56],"this":[57],"gap,":[58],"we":[59],"evaluate":[60],"six":[61],"methods":[63],"on":[64],"four":[65],"LLMs":[66],"across":[67],"two":[68],"jailbreak":[69],"benchmarks":[70],"three":[72],"settings:":[73],"Static,":[74],"Auto-DAN,":[75],"DSN.We":[77],"find":[78],"inflate":[82],"judge-based":[83],"attack":[84],"success":[85],"rate,":[86],"denoted":[87],"ASR,":[88],"jailbreaking,":[90],"largest":[93],"effects":[94],"appearing":[95],"in":[96],"biased":[97],"schemes":[98],"perturb":[100],"logits.At":[101],"same":[103],"time,":[104],"these":[105],"ASR":[106],"increases":[107],"often":[108],"do":[109],"not":[110],"reflect":[111],"higher":[112],"harmful-goal":[113,136],"compliance":[114],"when":[115],"measured":[116],"by":[117,120],"StrongREJECT":[118],"or":[119],"human":[121,145],"judgments.This":[122],"suggests":[123],"ASRonly":[125],"evaluations":[126],"be":[128],"brittle":[129],"to":[130],"decoding":[131],"perturbations":[132],"overestimate":[135],"compliance,":[137],"motivating":[138],"complementary":[139],"goalcompliance":[140],"metrics":[141],"(e.g.,":[142],"StrongREJECT)":[143],"evaluations.":[146]},"counts_by_year":[],"updated_date":"2026-10-28T10:33:24.094632","created_date":"2026-07-02T00:00:00"}