{"id":"https://openalex.org/W7220154847","doi":null,"title":"Continual Guardrail Learning for Tool-Using LLM Agents: Cross-Benchmark Jailbreak Detection, Indirect Prompt-Injection Filtering, and Utility Preservation","display_name":"Continual Guardrail Learning for Tool-Using LLM Agents: Cross-Benchmark Jailbreak Detection, Indirect Prompt-Injection Filtering, and Utility Preservation","publication_year":2026,"publication_date":"2026-08-07","ids":{"openalex":"https://openalex.org/W7220154847"},"language":"en","primary_location":{"id":"pmh:oai:journal.jci.co.id:article/590","doi":null,"is_oa":true,"landing_page_url":"https://journal.jci.co.id/index/article/view/590","pdf_url":null,"source":{"id":"https://openalex.org/S7407075622","display_name":"Journal of Information Systems and Business Technology","issn_l":"3109-8886","issn":["3109-8886"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"3109-8886","raw_type":"info:eu-repo/semantics/article"},"volume":null,"type":"article","is_traditional":true,"indexed_in":[],"open_access":{"is_oa":true,"oa_status":"diamond","oa_url":"https://journal.jci.co.id/index/article/view/590","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5156913709","display_name":"Hao Ran","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ran, Hao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5117707621","display_name":"NATALIE FOSTER","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Foster, Natalie","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5143457282","display_name":"Bo Liang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liang, Bo","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5157289134","display_name":"Justin Meyer","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Meyer, Justin","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":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.68960614,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.9984999895095825,"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":0.9984999895095825,"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":0.00039999998989515007,"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":0.00019999999494757503,"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/jailbreak-detection","display_name":"jailbreak detection","score":0.8119999766349792},{"id":"https://openalex.org/keywords/prompt-injection","display_name":"prompt injection","score":0.8349999785423279},{"id":"https://openalex.org/keywords/continual-learning","display_name":"continual learning","score":0.5720000267028809},{"id":"https://openalex.org/keywords/utility-preservation","display_name":"utility preservation","score":0.5590000152587891},{"id":"https://openalex.org/keywords/agentdojo","display_name":"AgentDojo","score":0.6380000114440918},{"id":"https://openalex.org/keywords/gpt-4o","display_name":"GPT-4o","score":0.671999990940094}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6662999987602234},{"id":"https://openalex.org/C32587265","wikidata":"https://www.wikidata.org/wiki/Q1182260","display_name":"Data deduplication","level":2,"score":0.5113000273704529},{"id":"https://openalex.org/C95623464","wikidata":"https://www.wikidata.org/wiki/Q1096149","display_name":"Classifier (UML)","level":2,"score":0.5098999738693237},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4611999988555908},{"id":"https://openalex.org/C43521106","wikidata":"https://www.wikidata.org/wiki/Q2165493","display_name":"Pipeline (software)","level":2,"score":0.4510999917984009},{"id":"https://openalex.org/C46686674","wikidata":"https://www.wikidata.org/wiki/Q466303","display_name":"Boosting (machine learning)","level":2,"score":0.4302999973297119},{"id":"https://openalex.org/C111065885","wikidata":"https://www.wikidata.org/wiki/Q1189053","display_name":"Fuzz testing","level":3,"score":0.40610000491142273},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.40549999475479126},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.37950000166893005},{"id":"https://openalex.org/C100660578","wikidata":"https://www.wikidata.org/wiki/Q18733","display_name":"Recall","level":2,"score":0.37709999084472656},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.36910000443458557},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.3677999973297119},{"id":"https://openalex.org/C2779777834","wikidata":"https://www.wikidata.org/wiki/Q4202277","display_name":"Enforcement","level":2,"score":0.367000013589859},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.35199999809265137},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.31700000166893005},{"id":"https://openalex.org/C153083717","wikidata":"https://www.wikidata.org/wiki/Q6535263","display_name":"Leverage (statistics)","level":2,"score":0.2833999991416931},{"id":"https://openalex.org/C81669768","wikidata":"https://www.wikidata.org/wiki/Q2359161","display_name":"Precision and recall","level":2,"score":0.2680000066757202},{"id":"https://openalex.org/C127705205","wikidata":"https://www.wikidata.org/wiki/Q5748245","display_name":"Heuristics","level":2,"score":0.25450000166893005},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.2538999915122986},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.25110000371932983}],"mesh":[],"locations_count":1,"locations":[{"id":"pmh:oai:journal.jci.co.id:article/590","doi":null,"is_oa":true,"landing_page_url":"https://journal.jci.co.id/index/article/view/590","pdf_url":null,"source":{"id":"https://openalex.org/S7407075622","display_name":"Journal of Information Systems and Business Technology","issn_l":"3109-8886","issn":["3109-8886"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"3109-8886","raw_type":"info:eu-repo/semantics/article"}],"best_oa_location":{"id":"pmh:oai:journal.jci.co.id:article/590","doi":null,"is_oa":true,"landing_page_url":"https://journal.jci.co.id/index/article/view/590","pdf_url":null,"source":{"id":"https://openalex.org/S7407075622","display_name":"Journal of Information Systems and Business Technology","issn_l":"3109-8886","issn":["3109-8886"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"3109-8886","raw_type":"info:eu-repo/semantics/article"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W7172000952","https://openalex.org/W7223512052","https://openalex.org/W7220167706","https://openalex.org/W7172108453","https://openalex.org/W7204931828","https://openalex.org/W7222349432","https://openalex.org/W7165183485","https://openalex.org/W7167825513","https://openalex.org/W7204584025","https://openalex.org/W7166886473"],"abstract_inverted_index":{"Tool-using":[0],"large":[1],"language":[2],"model":[3],"agents":[4],"process":[5],"untrusted":[6],"text":[7,32],"while":[8,26,152,170],"holding":[9],"permissions":[10],"to":[11,73,113,117,150,158],"communicate,":[12],"book":[13],"services,":[14],"and":[15,47,61,70,76,97,119,131],"move":[16],"funds.":[17],"This":[18],"study":[19,33],"evaluates":[20],"lightweight":[21],"guardrails":[22,161],"under":[23,168],"changing":[24],"attacks":[25,60,139],"preserving":[27],"authorized":[28],"task":[29],"utility.":[30],"The":[31],"uses":[34],"all":[35],"6,899":[36],"saved":[37],"runs":[38],"in":[39,121],"one":[40],"internally":[41],"consistent":[42],"AgentDojo":[43,81,116,138],"v1":[44],"GPT-4o":[45],"pipeline":[46],"200":[48],"JailbreakBench":[49,118],"prompts":[50],"covering":[51],"100":[52],"behavior":[53],"groups.":[54],"Deduplication":[55],"produced":[56],"5,280":[57],"texts:":[58],"1,450":[59],"3,830":[62],"benign":[63,94],"records.":[64],"Injection-goal":[65],"grouping":[66],"assigned":[67],"3,169,":[68],"1,056,":[69],"1,055":[71],"records":[72],"training,":[74],"validation,":[75],"testing":[77],"without":[78],"sharing":[79],"an":[80],"attack":[82,146],"target.":[83],"A":[84,100],"normalized":[85],"word-character":[86],"classifier":[87],"achieved":[88],"85.43%":[89],"F1,":[90],"90.31%":[91],"recall,":[92],"7.96%":[93],"false-positive":[95],"rate,":[96],"0.9613":[98],"ROC-AUC.":[99],"1,000-round":[101],"group":[102],"bootstrap":[103],"gave":[104],"a":[105],"wide":[106],"59.71-97.68%":[107],"F1":[108],"interval.":[109],"Cross-source":[110],"recall":[111],"fell":[112],"33.00%":[114],"from":[115,148,156],"0.16%":[120],"the":[122,174],"reverse":[123],"direction.":[124],"Sequential":[125],"tests":[126],"compared":[127],"frozen,":[128],"benign-anchored,":[129],"hard-replay,":[130],"soft-label-replay":[132],"learners.":[133],"Reanalysis":[134],"of":[135],"629":[136],"matched":[137],"showed":[140],"that":[141],"tool":[142],"filtering":[143],"reduced":[144],"targeted":[145],"success":[147],"47.69%":[149],"6.84%":[151],"increasing":[153],"safe":[154],"utility":[155],"29.73%":[157],"52.62%.":[159],"Layered":[160],"are":[162],"therefore":[163],"necessary:":[164],"replay":[165],"maintains":[166],"coverage":[167],"shift,":[169],"tool-level":[171],"enforcement":[172],"provided":[173],"strongest":[175],"observed":[176],"end-to-end":[177],"security-utility":[178],"balance.":[179]},"counts_by_year":[],"updated_date":"2026-10-28T10:33:24.094632","created_date":"2026-10-05T00:00:00"}