{"id":"https://openalex.org/W4407131685","doi":"https://doi.org/10.18653/v1/2026.findings-acl.566","title":"Peering Behind the Shield: Guardrail Identification in Large Language Models","display_name":"Peering Behind the Shield: Guardrail Identification in Large Language Models","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W4407131685","doi":"https://doi.org/10.18653/v1/2026.findings-acl.566"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2026.findings-acl.566","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.566","pdf_url":"https://aclanthology.org/2026.findings-acl.566.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 2026","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2026.findings-acl.566.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5001241589","display_name":"Ziqing Yang","orcid":"https://orcid.org/0000-0002-0492-0862"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ziqing Yang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102863764","display_name":"Yixin Wu","orcid":"https://orcid.org/0000-0002-3000-9423"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yixin Wu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101540234","display_name":"Rui Wen","orcid":"https://orcid.org/0000-0001-7728-1873"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Rui Wen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5004014878","display_name":"Michael Backes","orcid":"https://orcid.org/0000-0002-9326-6400"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Michael Backes","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":null,"display_name":"Yang Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yang Zhang","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":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.00516012,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"11663","last_page":"11676"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.9341999888420105,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.9341999888420105,"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/peering","display_name":"Peering","score":0.9569084644317627},{"id":"https://openalex.org/keywords/shield","display_name":"Shield","score":0.7570128440856934},{"id":"https://openalex.org/keywords/identification","display_name":"Identification (biology)","score":0.6910563111305237},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.4356442987918854},{"id":"https://openalex.org/keywords/geology","display_name":"Geology","score":0.3567385673522949},{"id":"https://openalex.org/keywords/history","display_name":"History","score":0.33336925506591797},{"id":"https://openalex.org/keywords/world-wide-web","display_name":"World Wide Web","score":0.12129810452461243},{"id":"https://openalex.org/keywords/petrology","display_name":"Petrology","score":0.10600784420967102}],"concepts":[{"id":"https://openalex.org/C95981142","wikidata":"https://www.wikidata.org/wiki/Q772532","display_name":"Peering","level":3,"score":0.9569084644317627},{"id":"https://openalex.org/C138081364","wikidata":"https://www.wikidata.org/wiki/Q852013","display_name":"Shield","level":2,"score":0.7570128440856934},{"id":"https://openalex.org/C116834253","wikidata":"https://www.wikidata.org/wiki/Q2039217","display_name":"Identification (biology)","level":2,"score":0.6910563111305237},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.4356442987918854},{"id":"https://openalex.org/C127313418","wikidata":"https://www.wikidata.org/wiki/Q1069","display_name":"Geology","level":0,"score":0.3567385673522949},{"id":"https://openalex.org/C95457728","wikidata":"https://www.wikidata.org/wiki/Q309","display_name":"History","level":0,"score":0.33336925506591797},{"id":"https://openalex.org/C136764020","wikidata":"https://www.wikidata.org/wiki/Q466","display_name":"World Wide Web","level":1,"score":0.12129810452461243},{"id":"https://openalex.org/C5900021","wikidata":"https://www.wikidata.org/wiki/Q163082","display_name":"Petrology","level":1,"score":0.10600784420967102},{"id":"https://openalex.org/C59822182","wikidata":"https://www.wikidata.org/wiki/Q441","display_name":"Botany","level":1,"score":0.0},{"id":"https://openalex.org/C86803240","wikidata":"https://www.wikidata.org/wiki/Q420","display_name":"Biology","level":0,"score":0.0},{"id":"https://openalex.org/C110875604","wikidata":"https://www.wikidata.org/wiki/Q75","display_name":"The Internet","level":2,"score":0.0}],"mesh":[],"locations_count":7,"locations":[{"id":"doi:10.18653/v1/2026.findings-acl.566","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.566","pdf_url":"https://aclanthology.org/2026.findings-acl.566.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 2026","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2502.01241","is_oa":true,"landing_page_url":"https://arxiv.org/abs/2502.01241","pdf_url":"https://arxiv.org/pdf/2502.01241","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,"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":"pmh:oai:figshare.com:article/32775144","is_oa":true,"landing_page_url":"https://figshare.com/articles/preprint/Peering_Behind_the_Shield_Guardrail_Identification_in_Large_Language_Models/32775144","pdf_url":null,"source":{"id":"https://openalex.org/S4377196282","display_name":"Figshare","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I4210132348","host_organization_name":"Figshare (United Kingdom)","host_organization_lineage":["https://openalex.org/I4210132348"],"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.2502.01241","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2502.01241","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,"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":null,"raw_source_name":null,"raw_type":"Preprint"},{"id":"doi:10.60882/cispa.32775144","is_oa":true,"landing_page_url":"https://doi.org/10.60882/cispa.32775144","pdf_url":null,"source":{"id":"https://openalex.org/S7407050916","display_name":"CISPA Helmholtz Center","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"},{"id":"doi:10.60882/cispa.32775144.v1","is_oa":true,"landing_page_url":"https://doi.org/10.60882/cispa.32775144.v1","pdf_url":null,"source":{"id":"https://openalex.org/S7407050916","display_name":"CISPA Helmholtz Center","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"},{"id":"doi:10.60882/cispa.32775144.v2","is_oa":true,"landing_page_url":"https://doi.org/10.60882/cispa.32775144.v2","pdf_url":null,"source":{"id":"https://openalex.org/S7407050916","display_name":"CISPA Helmholtz Center","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.findings-acl.566","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.566","pdf_url":"https://aclanthology.org/2026.findings-acl.566.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 2026","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4407131685.pdf","grobid_xml":"https://content.openalex.org/works/W4407131685.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W3110524642","https://openalex.org/W3165128667","https://openalex.org/W4281924170","https://openalex.org/W410534951","https://openalex.org/W3139861635","https://openalex.org/W2180109163","https://openalex.org/W3139924076","https://openalex.org/W2068704438","https://openalex.org/W1545439239","https://openalex.org/W2528595985"],"abstract_inverted_index":{"With":[0],"the":[1,32,44,65,82,139],"rapid":[2],"adoption":[3],"of":[4,67,84,95,141],"large":[5],"language":[6],"models":[7],"(LLMs),":[8],"conversational":[9],"AI":[10,72,155],"agents":[11,22,121],"have":[12],"become":[13],"widely":[14],"deployed":[15,69],"across":[16,119],"real-world":[17,154],"applications.To":[18],"enhance":[19],"safety,":[20],"these":[21],"are":[23],"often":[24],"equipped":[25],"with":[26],"guardrails":[27,33,68,125],"that":[28,58,127],"moderate":[29],"harmful":[30],"content.Identifying":[31],"in":[34,70,78,133,153],"an":[35],"agent":[36],"thus":[37],"becomes":[38],"critical":[39],"for":[40],"adversaries":[41],"to":[42,63,115],"understand":[43],"system":[45],"and":[46,87,105,109,122],"design":[47],"guardspecific":[48,60],"attacks.In":[49],"this":[50,79],"work,":[51],"we":[52],"introduce":[53],"AP-Test,":[54],"a":[55,93,110,147],"novel":[56],"approach":[57],"leverages":[59],"adversarial":[61],"prompts":[62],"detect":[64],"identity":[66],"black-box":[71],"agents.Our":[73],"method":[74],"addresses":[75],"key":[76],"challenges":[77],"task,":[80],"including":[81],"influence":[83],"safetyaligned":[85],"LLMs":[86],"other":[88],"guardrails,":[89],"as":[90,92],"well":[91],"lack":[94],"principled":[96],"decision-making":[97],"strategies.AP-Test":[98],"employs":[99],"two":[100],"complementary":[101],"testing":[102],"strategies,":[103],"input":[104],"output":[106],"guard":[107],"tests,":[108],"new":[111],"metric,":[112],"match":[113],"score,":[114],"enable":[116],"robust":[117],"identification.Experiments":[118],"diverse":[120],"four":[123],"open-source":[124],"demonstrate":[126],"AP-Test":[128],"achieves":[129],"perfect":[130],"classification":[131],"accuracy":[132],"multiple":[134],"scenarios.Ablation":[135],"studies":[136],"further":[137],"highlight":[138],"necessity":[140],"our":[142],"proposed":[143],"components.Our":[144],"findings":[145],"reveal":[146],"practical":[148],"path":[149],"toward":[150],"guardrail":[151],"identification":[152],"systems.":[156]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2025-10-10T00:00:00"}