{"id":"https://openalex.org/W7158209606","doi":"https://doi.org/10.1145/3805621.3807629","title":"Towards Graph-Based Detection of Jailbreak and Prompt-Leakage Attacks in LLMs","display_name":"Towards Graph-Based Detection of Jailbreak and Prompt-Leakage Attacks in LLMs","publication_year":2026,"publication_date":"2026-04-27","ids":{"openalex":"https://openalex.org/W7158209606","doi":"https://doi.org/10.1145/3805621.3807629"},"language":null,"primary_location":{"id":"doi:10.1145/3805621.3807629","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3805621.3807629","pdf_url":null,"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 Sixth European Workshop on Machine Learning and Systems","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://doi.org/10.1145/3805621.3807629","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5085156523","display_name":"Javad Forough","orcid":"https://orcid.org/0000-0003-3399-2440"},"institutions":[{"id":"https://openalex.org/I47508984","display_name":"Imperial College London","ror":"https://ror.org/041kmwe10","country_code":"GB","type":"education","lineage":["https://openalex.org/I47508984"]}],"countries":["GB"],"is_corresponding":true,"raw_author_name":"Javad Forough","raw_affiliation_strings":["Department of Computing, Imperial College London, London, London, United Kingdom"],"raw_orcid":"https://orcid.org/0000-0003-3399-2440","affiliations":[{"raw_affiliation_string":"Department of Computing, Imperial College London, London, London, United Kingdom","institution_ids":["https://openalex.org/I47508984"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5094348631","display_name":"Mohammad M Maheri","orcid":null},"institutions":[{"id":"https://openalex.org/I47508984","display_name":"Imperial College London","ror":"https://ror.org/041kmwe10","country_code":"GB","type":"education","lineage":["https://openalex.org/I47508984"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Mohammad M Maheri","raw_affiliation_strings":["Department of Computing, Imperial College London, London, London, United Kingdom"],"raw_orcid":"https://orcid.org/0009-0005-1905-1611","affiliations":[{"raw_affiliation_string":"Department of Computing, Imperial College London, London, London, United Kingdom","institution_ids":["https://openalex.org/I47508984"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5134860591","display_name":"Hamed Haddadi","orcid":"https://orcid.org/0000-0002-5895-8903"},"institutions":[{"id":"https://openalex.org/I47508984","display_name":"Imperial College London","ror":"https://ror.org/041kmwe10","country_code":"GB","type":"education","lineage":["https://openalex.org/I47508984"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Hamed Haddadi","raw_affiliation_strings":["Department of Computing, Imperial College London, London, London, United Kingdom"],"raw_orcid":"https://orcid.org/0000-0002-5895-8903","affiliations":[{"raw_affiliation_string":"Department of Computing, Imperial College London, London, London, United Kingdom","institution_ids":["https://openalex.org/I47508984"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":3,"corresponding_author_ids":["https://openalex.org/A5085156523"],"corresponding_institution_ids":["https://openalex.org/I47508984"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.96249169,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"315","last_page":"322"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.901199996471405,"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":0.901199996471405,"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/T11424","display_name":"Security and Verification in Computing","score":0.03660000115633011,"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":0.009399999864399433,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/adversarial-system","display_name":"Adversarial system","score":0.8447999954223633},{"id":"https://openalex.org/keywords/compromise","display_name":"Compromise","score":0.641700029373169},{"id":"https://openalex.org/keywords/security-token","display_name":"Security token","score":0.599399983882904},{"id":"https://openalex.org/keywords/threat-model","display_name":"Threat model","score":0.41100001335144043},{"id":"https://openalex.org/keywords/graph","display_name":"Graph","score":0.4011000096797943},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.39489999413490295}],"concepts":[{"id":"https://openalex.org/C37736160","wikidata":"https://www.wikidata.org/wiki/Q1801315","display_name":"Adversarial system","level":2,"score":0.8447999954223633},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7120000123977661},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.6506999731063843},{"id":"https://openalex.org/C46355384","wikidata":"https://www.wikidata.org/wiki/Q726686","display_name":"Compromise","level":2,"score":0.641700029373169},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.599399983882904},{"id":"https://openalex.org/C140547941","wikidata":"https://www.wikidata.org/wiki/Q7797194","display_name":"Threat model","level":2,"score":0.41100001335144043},{"id":"https://openalex.org/C132525143","wikidata":"https://www.wikidata.org/wiki/Q141488","display_name":"Graph","level":2,"score":0.4011000096797943},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.39489999413490295},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3124000132083893},{"id":"https://openalex.org/C41065033","wikidata":"https://www.wikidata.org/wiki/Q2825412","display_name":"Adversary","level":2,"score":0.30809998512268066},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.29820001125335693},{"id":"https://openalex.org/C144986985","wikidata":"https://www.wikidata.org/wiki/Q871236","display_name":"Hierarchical database model","level":2,"score":0.2522999942302704}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3805621.3807629","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3805621.3807629","pdf_url":null,"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 Sixth European Workshop on Machine Learning and Systems","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.1145/3805621.3807629","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3805621.3807629","pdf_url":null,"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 Sixth European Workshop on Machine Learning and Systems","raw_type":"proceedings-article"},"sustainable_development_goals":[{"score":0.8379668593406677,"display_name":"Peace, Justice and strong institutions","id":"https://metadata.un.org/sdg/16"}],"awards":[{"id":"https://openalex.org/G3894894773","display_name":null,"funder_award_id":"EP/W005271/1","funder_id":"https://openalex.org/F4320314731","funder_display_name":"UK Research and Innovation"}],"funders":[{"id":"https://openalex.org/F4320314731","display_name":"UK Research and Innovation","ror":"https://ror.org/001aqnf71"}],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":4,"referenced_works":["https://openalex.org/W1546425147","https://openalex.org/W4388886073","https://openalex.org/W4405181744","https://openalex.org/W4405992904"],"related_works":[],"abstract_inverted_index":{"Large":[0],"Language":[1],"Models":[2],"(LLMs)":[3],"are":[4],"vulnerable":[5],"to":[6],"adversarial":[7,59],"input":[8,38,67],"prompts":[9,60],"that":[10,22,57,87],"compromise":[11],"both":[12],"safety":[13],"and":[14,19,40,76,83],"confidentiality,":[15],"including":[16],"policy-violation":[17],"jailbreaks":[18],"prompt-leakage":[20,84],"attacks":[21,29],"extract":[23],"private":[24],"system":[25],"instructions.":[26],"Although":[27],"these":[28],"pursue":[30],"different":[31],"objectives,":[32],"they":[33],"operate":[34],"entirely":[35],"at":[36],"the":[37,95],"level":[39],"often":[41],"share":[42],"structural":[43],"manipulation":[44],"patterns,":[45],"motivating":[46],"unified":[47],"pre-inference":[48,99],"defenses.":[49],"We":[50],"present":[51],"GuardNet,":[52],"a":[53,69],"hierarchical":[54],"graph-based":[55],"framework":[56],"detects":[58],"before":[61],"model":[62],"execution":[63],"by":[64],"representing":[65],"each":[66],"as":[68],"hybrid":[70],"token":[71],"graph":[72],"combining":[73],"sequential,":[74],"syntactic,":[75],"attention-derived":[77],"relations.":[78],"Experiments":[79],"on":[80],"representative":[81],"jailbreak":[82],"benchmarks":[85],"show":[86],"GuardNet":[88],"consistently":[89],"outperforms":[90],"existing":[91],"input-side":[92],"defenses,":[93],"demonstrating":[94],"effectiveness":[96],"of":[97],"structure-aware":[98],"protection":[100],"for":[101],"deployed":[102],"LLM":[103],"systems.":[104]},"counts_by_year":[],"updated_date":"2026-04-30T06:11:10.768123","created_date":"2026-04-30T00:00:00"}