{"id":"https://openalex.org/W4416549774","doi":"https://doi.org/10.18653/v1/2026.findings-acl.1300","title":"False Sense of Security: Why Probing-based Malicious Input Detection Fails to Generalize","display_name":"False Sense of Security: Why Probing-based Malicious Input Detection Fails to Generalize","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W4416549774","doi":"https://doi.org/10.18653/v1/2026.findings-acl.1300"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2026.findings-acl.1300","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1300","pdf_url":"https://aclanthology.org/2026.findings-acl.1300.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.1300.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5100417007","display_name":"Cheng Wang","orcid":"https://orcid.org/0000-0002-4752-0316"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Cheng Wang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5027049671","display_name":"Zeming Wei","orcid":"https://orcid.org/0000-0002-9867-3929"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zeming Wei","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100432506","display_name":"Qin Liu","orcid":"https://orcid.org/0000-0003-0719-148X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Qin Liu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Wenxuan Zhou","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wenxuan Zhou","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5102861481","display_name":"Muhao Chen","orcid":"https://orcid.org/0000-0003-0118-3147"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Muhao Chen","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":0.02742588,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"26100","last_page":"26113"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.4948999881744385,"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":0.4948999881744385,"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/T11241","display_name":"Advanced Malware Detection Techniques","score":0.12210000306367874,"subfield":{"id":"https://openalex.org/subfields/1711","display_name":"Signal Processing"},"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/T11644","display_name":"Spam and Phishing Detection","score":0.05260000005364418,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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/raising","display_name":"Raising (metalworking)","score":0.536899983882904},{"id":"https://openalex.org/keywords/simple","display_name":"Simple (philosophy)","score":0.4722000062465668},{"id":"https://openalex.org/keywords/semantics","display_name":"Semantics (computer science)","score":0.33709999918937683},{"id":"https://openalex.org/keywords/work","display_name":"Work (physics)","score":0.3149999976158142},{"id":"https://openalex.org/keywords/false-alarm","display_name":"False alarm","score":0.3003999888896942}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7113000154495239},{"id":"https://openalex.org/C2780589192","wikidata":"https://www.wikidata.org/wiki/Q7285140","display_name":"Raising (metalworking)","level":2,"score":0.536899983882904},{"id":"https://openalex.org/C2780586882","wikidata":"https://www.wikidata.org/wiki/Q7520643","display_name":"Simple (philosophy)","level":2,"score":0.4722000062465668},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4056999981403351},{"id":"https://openalex.org/C184337299","wikidata":"https://www.wikidata.org/wiki/Q1437428","display_name":"Semantics (computer science)","level":2,"score":0.33709999918937683},{"id":"https://openalex.org/C18762648","wikidata":"https://www.wikidata.org/wiki/Q42213","display_name":"Work (physics)","level":2,"score":0.3149999976158142},{"id":"https://openalex.org/C2776836416","wikidata":"https://www.wikidata.org/wiki/Q1364844","display_name":"False alarm","level":2,"score":0.3003999888896942},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.2833999991416931},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.27709999680519104},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.2702000141143799},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.2676999866962433},{"id":"https://openalex.org/C2779814899","wikidata":"https://www.wikidata.org/wiki/Q332880","display_name":"Common sense","level":2,"score":0.2615000009536743},{"id":"https://openalex.org/C61797465","wikidata":"https://www.wikidata.org/wiki/Q1188986","display_name":"Term (time)","level":2,"score":0.2590999901294708}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.18653/v1/2026.findings-acl.1300","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1300","pdf_url":"https://aclanthology.org/2026.findings-acl.1300.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:2509.03888","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2509.03888","pdf_url":"https://arxiv.org/pdf/2509.03888","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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2509.03888","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2509.03888","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":null,"license_id":null,"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.1300","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1300","pdf_url":"https://aclanthology.org/2026.findings-acl.1300.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":[{"id":"https://openalex.org/G1795802330","display_name":"CICI: UCSS: Enabling a Safe and Directive Multi-modal Foundation Model Ecosystem for Food Science Research","funder_award_id":"2531126","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G4706632707","display_name":"Proto-OKN Theme 1: Knowledge Graph Construction for Resilient, Trustworthy, and Secure Software Supply Chains","funder_award_id":"2333736","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"}],"funders":[{"id":"https://openalex.org/F4320306076","display_name":"National Science Foundation","ror":"https://ror.org/021nxhr62"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4416549774.pdf","grobid_xml":"https://content.openalex.org/works/W4416549774.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Large":[0],"Language":[1],"Models":[2],"(LLMs)":[3],"can":[4],"comply":[5],"with":[6,99],"harmful":[7],"instructions,":[8],"raising":[9],"serious":[10],"safety":[11,44],"concerns":[12],"despite":[13],"their":[14],"impressive":[15],"capabilities.Recent":[16],"work":[17],"has":[18],"leveraged":[19],"probing-based":[20],"approaches":[21,119],"to":[22,96,103,124],"study":[23],"the":[24,73,122,138],"separability":[25],"of":[26,92,106,114,140],"malicious":[27],"and":[28,35,71,79,120,128],"benign":[29],"inputs":[30],"in":[31,137,145],"LLMs'":[32],"internal":[33],"representations,":[34],"researchers":[36],"have":[37],"proposed":[38],"using":[39],"such":[40],"probing":[41],"methods":[42],"for":[43,131],"detection.We":[45],"systematically":[46],"re-examine":[47],"this":[48,69,146],"paradigm.Motivated":[49],"by":[50],"poor":[51],"out-of-distribution":[52],"performance,":[53],"we":[54,67,133],"hypothesize":[55],"that":[56],"probes":[57],"learn":[58],"superficial":[59],"patterns":[60,75,78],"rather":[61],"than":[62],"semantic":[63],"harmfulness.Through":[64],"controlled":[65,97],"experiments,":[66],"confirm":[68],"hypothesis":[70],"identify":[72],"specific":[74],"learned:":[76],"instructional":[77],"trigger":[80],"words.Our":[81],"investigation":[82],"follows":[83],"a":[84,111],"systematic":[85],"approach,":[86],"progressing":[87],"from":[88],"demonstrating":[89],"comparable":[90],"performance":[91],"simple":[93],"n-gram":[94],"methods,":[95],"experiments":[98],"semantically":[100],"cleaned":[101],"datasets,":[102],"detailed":[104],"analysis":[105],"pattern":[107],"dependencies.These":[108],"results":[109],"reveal":[110],"false":[112],"sense":[113],"security":[115],"around":[116],"current":[117],"probingbased":[118],"highlight":[121],"need":[123],"redesign":[125],"both":[126],"models":[127],"evaluation":[129],"protocols,":[130],"which":[132],"provide":[134],"further":[135,143],"discussions":[136],"hope":[139],"suggesting":[141],"responsible":[142],"research":[144],"direction.":[147]},"counts_by_year":[],"updated_date":"2026-08-05T07:39:15.569665","created_date":"2025-10-10T00:00:00"}