{"id":"https://openalex.org/W7162420762","doi":"https://doi.org/10.48550/arxiv.2605.24883","title":"Inverting the Shield: Systematically Generating Safety Tests from Policy Specifications","display_name":"Inverting the Shield: Systematically Generating Safety Tests from Policy Specifications","publication_year":2026,"publication_date":"2026-05-24","ids":{"openalex":"https://openalex.org/W7162420762","doi":"https://doi.org/10.48550/arxiv.2605.24883"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.24883","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.24883","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":false,"raw_source_name":null,"raw_type":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2605.24883","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5137027316","display_name":"Xiaoyue Lu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lu, Xiaoyue","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137069557","display_name":"Xianglin Yang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yang, Xianglin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137040563","display_name":"Haijun Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Haijun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137084734","display_name":"Jiahao Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Jiahao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5039612701","display_name":"Kuntai Cai","orcid":"https://orcid.org/0009-0006-8278-716X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Cai, Kuntai","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137058118","display_name":"Yan Xiao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xiao, Yan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5137025563","display_name":"Jin Song Dong","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Dong, Jin Song","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":null,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":null,"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":0.6518999934196472,"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":0.6518999934196472,"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/T10883","display_name":"Ethics and Social Impacts of AI","score":0.08879999816417694,"subfield":{"id":"https://openalex.org/subfields/3311","display_name":"Safety Research"},"field":{"id":"https://openalex.org/fields/33","display_name":"Social Sciences"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},{"id":"https://openalex.org/T13295","display_name":"Safety Systems Engineering in Autonomy","score":0.04230000078678131,"subfield":{"id":"https://openalex.org/subfields/2213","display_name":"Safety, Risk, Reliability and Quality"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/executable","display_name":"Executable","score":0.8303999900817871},{"id":"https://openalex.org/keywords/verifiable-secret-sharing","display_name":"Verifiable secret sharing","score":0.47850000858306885},{"id":"https://openalex.org/keywords/symbolic-execution","display_name":"Symbolic execution","score":0.45170000195503235},{"id":"https://openalex.org/keywords/domain","display_name":"Domain (mathematical analysis)","score":0.45089998841285706},{"id":"https://openalex.org/keywords/bridging","display_name":"Bridging (networking)","score":0.4424000084400177},{"id":"https://openalex.org/keywords/software","display_name":"Software","score":0.439300000667572},{"id":"https://openalex.org/keywords/code","display_name":"Code (set theory)","score":0.3303999900817871}],"concepts":[{"id":"https://openalex.org/C160145156","wikidata":"https://www.wikidata.org/wiki/Q778586","display_name":"Executable","level":2,"score":0.8303999900817871},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7228000164031982},{"id":"https://openalex.org/C115903868","wikidata":"https://www.wikidata.org/wiki/Q80993","display_name":"Software engineering","level":1,"score":0.6265000104904175},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.5121999979019165},{"id":"https://openalex.org/C85847156","wikidata":"https://www.wikidata.org/wiki/Q59015987","display_name":"Verifiable secret sharing","level":3,"score":0.47850000858306885},{"id":"https://openalex.org/C2779639559","wikidata":"https://www.wikidata.org/wiki/Q7661178","display_name":"Symbolic execution","level":3,"score":0.45170000195503235},{"id":"https://openalex.org/C36503486","wikidata":"https://www.wikidata.org/wiki/Q11235244","display_name":"Domain (mathematical analysis)","level":2,"score":0.45089998841285706},{"id":"https://openalex.org/C174348530","wikidata":"https://www.wikidata.org/wiki/Q188635","display_name":"Bridging (networking)","level":2,"score":0.4424000084400177},{"id":"https://openalex.org/C2777904410","wikidata":"https://www.wikidata.org/wiki/Q7397","display_name":"Software","level":2,"score":0.439300000667572},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.3303999900817871},{"id":"https://openalex.org/C111498074","wikidata":"https://www.wikidata.org/wiki/Q173326","display_name":"Formal verification","level":2,"score":0.311599999666214},{"id":"https://openalex.org/C80519477","wikidata":"https://www.wikidata.org/wiki/Q3532236","display_name":"Scenario testing","level":3,"score":0.30169999599456787},{"id":"https://openalex.org/C128942645","wikidata":"https://www.wikidata.org/wiki/Q1568346","display_name":"Test case","level":3,"score":0.29789999127388},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.2888000011444092},{"id":"https://openalex.org/C116253237","wikidata":"https://www.wikidata.org/wiki/Q1437424","display_name":"Formal specification","level":2,"score":0.28610000014305115},{"id":"https://openalex.org/C2777267654","wikidata":"https://www.wikidata.org/wiki/Q3519023","display_name":"Test (biology)","level":2,"score":0.28200000524520874},{"id":"https://openalex.org/C201677973","wikidata":"https://www.wikidata.org/wiki/Q1209840","display_name":"Specification language","level":2,"score":0.2786000072956085},{"id":"https://openalex.org/C2984968299","wikidata":"https://www.wikidata.org/wiki/Q1077784","display_name":"Software tool","level":3,"score":0.27059999108314514},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.2655999958515167},{"id":"https://openalex.org/C2984328558","wikidata":"https://www.wikidata.org/wiki/Q188522","display_name":"Software testing","level":3,"score":0.25760000944137573}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.24883","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.24883","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:10.48550/arxiv.2605.24883","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.24883","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":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[{"score":0.4792276620864868,"display_name":"Peace, Justice and strong institutions","id":"https://metadata.un.org/sdg/16"}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"The":[0],"widespread":[1],"integration":[2],"of":[3,73,107],"Large":[4],"Language":[5],"Models":[6],"(LLMs)":[7],"necessitates":[8],"rigorous":[9],"and":[10,53,98,143,155,168],"systematic":[11,51],"safety":[12,23,145],"evaluation.":[13],"Existing":[14],"paradigms":[15],"either":[16],"rely":[17],"on":[18,45],"constructed":[19],"benchmarks":[20],"to":[21,31,56,77,160,177,181],"assess":[22],"from":[24],"predefined":[25],"perspectives,":[26],"or":[27],"employ":[28],"dynamic":[29],"red-teaming":[30],"probe":[32],"potential":[33],"vulnerabilities.":[34],"While":[35],"effective,":[36],"these":[37,61],"approaches":[38],"face":[39],"challenges,":[40],"as":[41,119],"they":[42],"depend":[43],"heavily":[44],"expert":[46],"domain":[47],"knowledge,":[48],"offer":[49],"limited":[50],"guarantees,":[52],"are":[54,117,133],"vulnerable":[55],"rapid":[57],"obsolescence.":[58],"To":[59],"address":[60],"limitations,":[62],"we":[63],"introduce":[64],"a":[65,92,108,173],"novel":[66],"framework":[67],"POLARIS":[68,80,127,150,171],"that":[69,149],"brings":[70],"the":[71,105],"rigor":[72],"specification-based":[74],"software":[75],"testing":[76],"AI":[78,169],"safety.":[79],"first":[81],"compiles":[82],"unstructured":[83],"natural-language":[84,138],"policies":[85,183],"into":[86,136],"First-Order":[87],"Logic":[88],"(FOL)":[89],"representations,":[90],"establishing":[91],"traceable":[93],"link":[94],"between":[95],"high-level":[96],"rules":[97],"concrete":[99],"test":[100,139],"cases.":[101],"This":[102],"formalization":[103],"enables":[104],"construction":[106],"Semantic":[109],"Policy":[110],"Graph,":[111],"where":[112],"complex":[113],"policy":[114,153],"violation":[115,130],"scenarios":[116],"encoded":[118],"traversable":[120],"paths.":[121],"By":[122],"systematically":[123],"exploring":[124],"this":[125],"graph,":[126],"uncovers":[128],"compositional":[129],"patterns,":[131],"which":[132],"then":[134],"instantiated":[135],"executable":[137],"queries,":[140],"enabling":[141],"coverage-driven":[142],"reproducible":[144],"testing.":[146],"Experiments":[147],"demonstrate":[148],"achieves":[151],"higher":[152],"coverage":[154],"attack":[156],"success":[157],"counts":[158],"compared":[159],"established":[161],"baselines.":[162],"Crucially,":[163],"by":[164],"bridging":[165],"formal":[166],"methods":[167],"safety,":[170],"provides":[172],"principled,":[174],"automated":[175],"approach":[176],"ensuring":[178],"LLMs":[179],"adhere":[180],"safety-critical":[182],"with":[184],"verifiable":[185],"traceability.":[186],"We":[187],"release":[188],"our":[189],"code":[190],"at":[191],"https://github.com/huac-lxy/POLARIS.":[192]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-05-27T00:00:00"}