{"id":"https://openalex.org/W7140126080","doi":"https://doi.org/10.18653/v1/2026.findings-eacl.198","title":"BitBypass: A New Direction in Jailbreaking Aligned Large Language Models with Bitstream Camouflage","display_name":"BitBypass: A New Direction in Jailbreaking Aligned Large Language Models with Bitstream Camouflage","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7140126080","doi":"https://doi.org/10.18653/v1/2026.findings-eacl.198"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.findings-eacl.198","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-eacl.198","pdf_url":"https://aclanthology.org/2026.findings-eacl.198.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: EACL 2026","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2026.findings-eacl.198.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":null,"display_name":"Kalyan Nakka","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kalyan Nakka","raw_affiliation_strings":["SPIES Research Lab , Dept. of CSE , Texas A&M University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"SPIES Research Lab , Dept. of CSE , Texas A&M University","institution_ids":[]}]},{"author_position":"last","author":{"id":null,"display_name":"Nitesh Saxena","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Nitesh Saxena","raw_affiliation_strings":["SPIES Research Lab , Dept. of CSE , Texas A&M University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"SPIES Research Lab , Dept. of CSE , Texas A&M University","institution_ids":[]}]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":2,"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":0.35536901,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"3808","last_page":"3834"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.24690000712871552,"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"}},"topics":[{"id":"https://openalex.org/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.24690000712871552,"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/T12380","display_name":"Authorship Attribution and Profiling","score":0.10980000346899033,"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/T12262","display_name":"Hate Speech and Cyberbullying Detection","score":0.10170000046491623,"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/camouflage","display_name":"Camouflage","score":0.7037000060081482},{"id":"https://openalex.org/keywords/bitstream","display_name":"Bitstream","score":0.38109999895095825},{"id":"https://openalex.org/keywords/semantics","display_name":"Semantics (computer science)","score":0.2800999879837036},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.2574999928474426}],"concepts":[{"id":"https://openalex.org/C2776196576","wikidata":"https://www.wikidata.org/wiki/Q196113","display_name":"Camouflage","level":2,"score":0.7037000060081482},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6848000288009644},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4948999881744385},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.45750001072883606},{"id":"https://openalex.org/C121684516","wikidata":"https://www.wikidata.org/wiki/Q7600677","display_name":"Computer graphics (images)","level":1,"score":0.4092999994754791},{"id":"https://openalex.org/C136695289","wikidata":"https://www.wikidata.org/wiki/Q415568","display_name":"Bitstream","level":3,"score":0.38109999895095825},{"id":"https://openalex.org/C62649853","wikidata":"https://www.wikidata.org/wiki/Q199687","display_name":"Remote sensing","level":1,"score":0.29170000553131104},{"id":"https://openalex.org/C184337299","wikidata":"https://www.wikidata.org/wiki/Q1437428","display_name":"Semantics (computer science)","level":2,"score":0.2800999879837036},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.2574999928474426},{"id":"https://openalex.org/C36464697","wikidata":"https://www.wikidata.org/wiki/Q451553","display_name":"Visualization","level":2,"score":0.243599995970726}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2026.findings-eacl.198","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-eacl.198","pdf_url":"https://aclanthology.org/2026.findings-eacl.198.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: EACL 2026","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.findings-eacl.198","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-eacl.198","pdf_url":"https://aclanthology.org/2026.findings-eacl.198.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: EACL 2026","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G4131819877","display_name":"Collaborative Research: SaTC: TTP: Medium: Intrusion-Tolerant Outsourced Storage for Cyber-Infrastructure","funder_award_id":"2201465","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G424997876","display_name":null,"funder_award_id":"OAC-2139358","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G7846119105","display_name":null,"funder_award_id":"CNS-2201465","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G8377161673","display_name":"CICI: UCSS: Towards Secure and Usable Push Notification Authentication for Collaborative Scientific Infrastructures","funder_award_id":"2139358","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/W7140126080.pdf","grobid_xml":"https://content.openalex.org/works/W7140126080.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"The":[0],"inherent":[1],"risk":[2],"of":[3,40,44,60,95,109,129,156,168],"generating":[4,140],"harmful":[5,141],"and":[6,31,57,121,136,142,158,166],"unsafe":[7,143],"content":[8],"by":[9,51,90],"Large":[10],"Language":[11],"Models":[12],"(LLMs),":[13],"has":[14],"highlighted":[15],"the":[16,37,42,61,127,164],"need":[17],"for":[18,35,80],"their":[19,133],"safety":[20,38,62,134],"alignment.Various":[21],"techniques":[22],"like":[23],"supervised":[24],"fine-tuning,":[25],"reinforcement":[26],"learning":[27],"from":[28],"human":[29],"feedback,":[30],"red-teaming":[32],"were":[33],"developed":[34],"ensuring":[36],"alignment":[39,135],"LLMs.However,":[41],"robustness":[43],"these":[45,161,172],"aligned":[46,82],"LLMs":[47],"is":[48],"always":[49],"challenged":[50],"adversarial":[52,106,124],"attacks":[53,153],"that":[54,75,147],"exploit":[55],"unexplored":[56],"underlying":[58],"vulnerabilities":[59],"alignment.In":[63],"this":[64],"paper,":[65],"we":[66,145],"develop":[67],"a":[68,85],"novel":[69],"black-box":[70],"jailbreak":[71,152],"attack,":[72],"called":[73],"BitBypass,":[74],"leverages":[76],"hyphen-separated":[77],"bitstream":[78],"camouflage":[79],"jailbreaking":[81,89,171],"LLMs.This":[83],"represents":[84],"new":[86],"direction":[87],"in":[88,123,131,154,170],"exploiting":[91],"fundamental":[92],"information":[93],"representation":[94],"data":[96],"as":[97],"continuous":[98],"bits,":[99],"rather":[100],"than":[101],"leveraging":[102],"prompt":[103],"engineering":[104],"or":[105],"manipulations.Our":[107],"evaluation":[108],"five":[110],"state-of-the-art":[111,151,173],"LLMs,":[112],"namely":[113],"GPT-4o,":[114],"Gemini":[115],"1.5,":[116],"Claude":[117],"3.5,":[118],"Llama":[119],"3.1,":[120],"Mixtral,":[122],"perspective,":[125],"revealed":[126],"capabilities":[128],"BitBypass":[130,148,169],"bypassing":[132],"tricking":[137],"them":[138],"into":[139],"content.Further,":[144],"observed":[146],"outperforms":[149],"several":[150],"terms":[155],"stealthiness":[157],"attack":[159],"success.Overall,":[160],"results":[162],"highlights":[163],"effectiveness":[165],"efficiency":[167],"LLMs.":[174]},"counts_by_year":[],"updated_date":"2026-06-22T08:00:12.763002","created_date":"2026-03-07T00:00:00"}