AI时代智慧城市安全保障白皮书：技术框架与细分行业分析（卷二）-

AFCEA

SECURINGSMARTCITIESIN

THEAGEOFAI

VOLUME2:TECHNICALFRAMEWORKSANDSECTOR-SPECIFICANALYSIS

AnAFCEACybersecurityCommitteeand

HomelandSecurityCommitteeWhitePaper(2026)

JANUARY2026

2

PRIMARYAUTHORSANDCONTRIBUTORS

Thiswhitepaperwasdevelopedbythefollowingprimaryauthorsandcontributors.Leadauthorssynthesizedtheresearch,conductedthecoreanalysisanddraftedtheprincipalsections;contributorsprovidedsourcematerials,suggestedanalyticalmodels,identifiedkeyareasfordeeperanalysisandstrengthenedthefinalrecommendationsthroughtargetedreviewsandedits.Affiliationsarelistedforidentificationonly,andnamesappearalphabetically,withleadauthorsnoted.

LeadAuthor

•Dr.RhondaFarrell,AFCEACybersecurityCommittee

ContributingAuthors

AFCEAInternational

•KhalilahFilmore,HomelandSecurityCommittee

•MarcusSachs,CybersecurityCommittee

•KathySwacina,HomelandSecurityCommittee,HISPIProjectCerebellum

•PaulWertz,HomelandSecurityCommittee

AmericanSocietyforQuality(ASQ)

•BrianDeniese,ASQGovernmentDivision,ProvinceofOntaria-RegionalMunicipalityofPeel

•MichaelMladjenovic,ASQStatisticsDivision

HolisticInformationSecurityPractitioner(HISP)Institute(HISPI)

•MarinaAlvarez,HISPIProjectCerebellumAdvocacyWorkingGroupCo-Lead

•TaiyeLambo,FounderHISPIandProjectCerebellum

•WilliamMcBorrough,HISPIProjectCerebellumPartnershipWorkingGroupLeadInstituteofElectricalandElectronicsEngineers(IEEE)

•AngeloFerraro,ArtificialIntelligenceCoalition,UniversityofSouthCarolina

Iridius.AI

•AlistairLowe-Norris,ChiefResponsibleAIOfficer

3

CONTENTS

PrimaryAuthorsandContributors 2

ExecutiveSummary 7

Background 9

1.0Introduction:AIinUrbanContexts:RisksandImperatives 11

2.0EvolvingSmartCityThreatLandscape 14

2.1ExpandingCyber-PhysicalRisk 16

2.2ComplexityandShadowAI 16

2.3GovernanceandTrust 17

2.4Conclusion:SecuringTrustintheAgeofIntelligentInfrastructure 17

3.0AIIntegrationAcrossCriticalInfrastructure 18

3.1AHolisticView:AIasaCross-SectorForce 18

3.2SectoralIntegrationandSharedRisks 19

3.3GovernanceImperativesAcrossSectors 20

3.4IntegrativeTable:AIApplicationsAcrossCriticalInfrastructure 21

3.5StructureofCriticalInfrastructureSectoralAnalysis 22

3.6Conclusion:ATippingPointforSmartCities 23

4.0CyberGovernanceGapsandChallenges 24

4.1FragmentedOversight 24

4.2GapsinRiskManagement 24

4.3InadequateTestingandEvaluation 25

4.4TransparencyandTrustDeficits 25

4.5Conclusion:TowardIntegrated,AdaptiveAIGovernance 25

5.0PolicyImperativesforSecureUrbanAI 26

5.1InstitutionalizeAIRiskGovernance 26

4

CONTENTS

5.2RedesignProcurementStandards 27

5.3EmbedEthicalandPrivacySafeguards 27

5.4LaunchCross-SectorTestBeds 27

5.5InvestinWorkforceDevelopment 28

5.6Conclusion:BuildingaSecureAIFutureforCities 28

6.0CaseStudiesandUseCases:AIinActionAcrossCities 29

6.1DomesticCaseStudies(UnitedStates) 29

6.1.1Houston,TX–AI-Powered911Triage 29

6.1.2LosAngeles,CA–Real-TimeTransitOptimization 30

6.1.3Boston,MA–SmartWasteInfrastructure 30

6.2InternationalCaseStudies 30

6.2.1Singapore–EthicalAIinUrbanPlanning 30

6.2.2Barcelona,Spain–UrbanDigitalTwinsWithConsent 30

6.2.3Dubai,UAE–PredictivePolicingandAISurveillance 31

6.3Cross-CuttingThemesandGovernanceInsights 31

6.4Conclusion:FromDeploymenttoStewardship 31

7.0AIGovernanceandAssurance:BuildingaResilientOversightArchitecture 33

7.1GovernancePillarsforUrbanAISystems 33

7.2AssuranceMechanisms:FromTheorytoPractice 34

7.3IntegrationWithPolicyandProcurement 34

7.4Conclusion:AssuranceasaStrategicCivicAsset 35

8.0StrategicRecommendations:GoverningAIforUrbanResilienceandPublicTrust 36

8.1TheAIResilienceIndex(ARI) 36

8.2StructureoftheIndex 37

5

CONTENTS

8.3ScoringMethodology 37

8.4ApplicationsandIntegration 38

8.5CorrelationWithVolume1Recommendations 38

9.0Conclusion:BuildingTrustworthyIntelligenceIntotheSmartCitiesofTomorrow 39

9.1WhyActionIsUrgent 39

9.2ACallforSharedStewardship 39

9.3TheRoadAhead:ResiliencebyDesign,EquitybyDefault 40

9.4FinalCharge:AFutureThatDeservesUs 40

Appendices 41

AppendixA:Models&FrameworksReferenced

41

AppendixB:GlossaryofTerms(basedonOECD,NIST,ISO,priorAFCEAAIPaper) 42

AppendixC:EmergingThreatModalities 45

AppendixD:CriticalInfrastructureSectoralAnalysis 47

D.1TransportationSystems:NavigatingComplexityWithIntelligentMobility 47

D.2EnergySector:PoweringCitiesWithIntelligentandSecureAISystems 48

D.3EmergencyServices:TrustworthyAIforLife-CriticalResponse 50

D.4WaterandWastewaterSystems:SecuringtheFlowofIntelligentInfrastructure 52

D.5HealthCareandPublicHealth:SecuringIntelligenceinLife-CriticalSystems 54

D.6InformationTechnology:SecuringtheDigitalNervousSystemofSmartCities 56

D.7Communications:TheBackboneofAI-AugmentedUrbanResilience 59

D.8GovernmentFacilities:AI-EnabledGovernanceandSecurityinSmartCities 61

D.9FinancialServices:AI’sExpandingRoleinUrbanFiscalGovernanceandRisk

Management 63

D.10CommercialFacilities:AI-EnhancedUrbanExperience,SecurityandEconomic

Activation 65

6

CONTENTS

D.11CriticalManufacturing:AIattheNexusofIndustrialResilienceandUrban

EconomicSecurity 67

D.12EnvironmentalProtection:AIforUrbanSustainability,ResilienceandJustice 69

D.13DefenseIndustrialBase:AIattheNexusofUrbanSecurity,Innovationand

NationalDefense

71

D.14ChemicalSector:AI-DrivenSafetyinUrbanIndustrialZones 73

D.15Manufacturing(General):AIasaCatalystforInclusiveUrbanIndustrial

Growth 75

D.16Dams:AI-EnhancedRiskManagementforUrbanWaterResilience 77

AppendixE:SelectedTools,PilotsandFrameworksforSmartCityAIGovernance 79

AppendixF:BibliographyofRelevantMaterials 82

F.1AcademicandIndustryLiterature 82

F.2AIGovernanceFrameworksandStandards

82

F.3Cross-CuttingFrameworksandToolKits 83

F.4CybersecurityandResilience

83

F.5EthicsandHumanRights 84

F.6GovernmentandPublicSectorStrategy 84

F.7NationalPolicyandStrategyDocuments

84

F.8ProfessionalAssociationWhitePapers 85

F.9SmartCitiesandSectorUseCases 85

F.9Miscellaneous 86

AppendixG:QuoteRepository 88

AppendixH:AuthorandContributorOrganizationOverviews

89

7

EXECUTIVESUMMARY

Volume2ofSecuringSmartCitiesintheAgeofAIadvancesthemissionestablishedin

Volume1:

Strategy,GovernanceandRecommendations

(2025):tointegratecybersecurity,governanceandethicsintotheAI-drivendigitaltransformationofsmartcitiesandtheiraccompanyingcriticalinfrastructurecomponents.WhilethefirstvolumesetthestagebychroniclingtheevolutionofAI-drivenurbansystems,scopingtheassociatedthreatlandscapeandpromulgatingthecoresetofstrategicrecommendations,Volume2focusesonproviding:

•DetailedmappingsofAIgovernanceandcybersecuritycontrolsacrossthe16U.S.criticalinfrastructuresectors[34,38]

•Casestudiesofreal-worldsmart-citydeployments,examiningriskscenariosandrecoverymodels

•AcrosswalkaligningNISTAIRMF,ISO/IEC42001,ISO56000andU.S.federalframeworkstostrengtheninteroperability[9,10,21,23,24,28]

•Practicalassessmenttoolsandmaturitymodelsforevaluatingresilience,workforcereadinessandethicalAIcompliance[6,15,23,39]

TheglobalAIpolicyenvironmenthasshifteddramatically.TheEUAIActreachedfinaladoption;OECD’sHiroshimaAIProcess(HAIP)introducedcross-borderaccountabilityprotocols;theArtificialIntelligenceSafetyInstituteConsortium(

AISIC

)andNISTAIRMF1.0advancedassurancestandards;andISO/IEC42001:2025establishedthefirstinternationalmanagement-systemframeworkforAIgovernance.Collectively,theseinstrumentstransformthelandscapeenvisionedinearlierworksfromaspirationalalignmenttocodifiedexpectation.[9,10,35,65]

Volume2ofthiswhitepapermovesbeyondconceptualstrategytoalsooffer:

•Appliedimplementationcasereviewsacrossall16criticalinfrastructuresectors

•AnewlyformalizedAIResilienceIndex(ARI)withmeasurableKPIs

•UpdatedgovernanceandassurancemodelsreflectingNISTAIRMF2.0andOECDHAIPcriteria

•InvestmentandfinancingpathwaystosustainAI-enabledresiliencethrough2030Asin

Volume1

,ourguidingthesisremains:

“Smartcitiescannotbetrulysmartunlesstheyaresecure,accountableand

citizencentered.”—AFCEA

8

Thetransitionfromframeworktofieldrequiresthreeconditions:

1.InstitutionalMaturity:Governancestructuresthattranslatepolicyintopracticethroughconsistentmeasurement.

2.OperationalIntegration:ResilientAIsystemsembeddedintothedailyfunctionsofurbanlife.

3.PublicTrust:TransparentandauditableprocessesthatmakeAIasharedcivicasset,notablackbox.

Volume2capturesthenextphaseofthisjourney,howcitiesaredeploying,measuringandgoverningAIintherealworldandwhatstepsremaintoachieveethical,secureandsustainableurbanintelligence.

“Wemustgovernartificialintelligenceanddigitalinfrastructurenotjustwithcode,

butwithconscience.”—AdaptedfromOECD.AI,2024

9

BACKGROUND

Volume1

establishedthatAI-enabledsmartcitiesfacearapidlyevolvingandmultilayeredthreatenvironment,whereadversariesexploittheconvergenceofinformationtechnology(IT)andoperationaltechnology(OT)systems,targetAImodelsandleverageautomatedattacktoolstoamplifyreal-worldimpact.Ithighlightedthatrisinginterdependenciesacrosscriticalinfrastructureheightenstheriskofcascadingfailures,whileregulatoryandgovernanceframeworksstruggletokeeppacewithAIadoption.CompoundingtheserisksaresignificantworkforceshortagesincyberandAIexpertise,limitingthepublicsector’scapacitytodefendandrespondeffectively.

Securingthefutureofsmartcitiesthusrequiresmorethantechnicalsafeguards;itdemandsvisionaryleadership,inclusivedesign,adaptiveregulationandcoordinatedinvestment.[3,5,10,14,16,18]Weurgepolicymakers,technologyleaders,cityplannersandcivilsocietytoactnow:collaborateacrossdomains,adoptsecure-by-designprinciples,prioritizeethicalAIgovernanceandensurethatsmartcitiesarenotonlyintelligent,butresilient,trustworthyandjust.

Thefindingsfrom

Volume1

underscoretheurgentneedforawhole-of-governmentandwhole-of-communitysecuritymodel,onethatunitestechnicalsafeguards,ethicalgovernance,publictrustandlong-termresilienceplanningastheessentialfoundationforfuturesmart-cityoperations.AscitiesacceleratetheiradoptionofAIacrosscriticalinfrastructure,publicservicesandurbanoperations,thisintegratedapproachbecomesevenmorevital.Volume2buildsontheseinsightsandunfoldsinthreeinterconnectedparts:first,byclarifyingthecoresecuritychallengesandsystemicrisksfacingAI-enabledsmartcities;next,bydefiningthegovernanceframeworks,ethicalsafeguardsandtechnicalsolutionsrequiredtomanageAIresponsiblyatscale;andfinally,byequippingseniorleaderswithapracticalblueprintforimplementation.Together,thesethreepartsprovideacoherentpathway,fromunderstandingtheevolvingthreatlandscape,toarchitectingresilientgovernance,toexecutingtrustworthyandfuture-readyAIsystems—ensuringthatsmartcitiescanadvanceinnovationwithoutcompromisingsafety,equityorpublictrust.

PartI:UnderstandingtheSmartCityAISecurityChallenge

Sections1–3establishthatAI-enabledsmartcitiesnowoperateoninvisible,always-onOT/ITinfrastructurethatamplifiesriskssuchasbias,cascadingfailuresandopaquedecision-making,requiringmoderngovernanceanchoredinframeworkslikeHISPIProjectCerebellumTAIM.Theyhighlightarapidlyevolvingthreatlandscapeshapedbylegacysystems,globalsupplychains,unevenoversightandemergingAIrisksacrossall16criticalinfrastructuresectors.Together,thesesectionsmakeclearthatsmartcitiescanonlysucceedwhenAIisgovernedaspublicinfrastructure,secure,accountableandcitizen-centered.[13,14,18,19,20,21,22,23,28,29,35,36,41,43,44,45,46,47,48,49,50,54,55]

10

PartII:Governance,EthicsandTechnicalSolutions

Sections4–6outlinethegovernance,policyandexecutionframeworksseniorleadersneedtosecureAI-enabledsmartcitiesatscale,diagnosingcurrentgapsinoversight,transparency,procurementandriskmanagement.Theytranslatethesegapsintoactionablereforms,includingethicalsafeguards,assurancemechanisms,lifecycle-alignedprocurement,cross-sectortestbedsandworkforcemodernization,whileillustratinghowleadingcitiesareapplyingAIacrosstransportation,publicsafety,energy,healthcareandemergencyresponse.Collectively,thesechaptersshifttheconversationfromdiagnosingvulnerabilitiestodefiningthestructural,policyandoperationalchangesrequiredfortrustworthy,resilientandmission-alignedurbanAIecosystems.[10-12,14,16,17,21,24,29,31,32,35,40,43-50,64,68]

PartIII:ActionandImplementation

Sections7–9provideseniorleaderswithahigh-levelblueprintforgoverningAI-enabledsmartcities,focusingonbuildingaresilientoversightarchitecturegroundedinaccountability,transparency,reliability,securityandprivacy.Theyoutlinetheessentialassurancemechanisms,governancestructures,workforcecapabilitiesandcross-sectorpartnershipsrequiredtooperationalizeresponsibleAIatscale,whileestablishingtoolssuchasAlgorithmicImpactAssessmentsandaSmartCityAIResilienceIndex.Together,thesesectionsmovefromstrategytoexecution,equippingexecutiveswithaclear,actionableframeworkfordeployingtrustworthy,resilientandfuture-readyurbanAIsystems.[6-17,19-23,30-32,40,55-57,64,68]

11

1.0INTRODUCTION:AIINURBAN

CONTEXTS:RISKSANDIMPERATIVES

Theriseofsmartcitiesmarksaprofoundshift:AIisnolongerperipheral;itisembeddedinthelifebloodofcivicinfrastructure.[10,34,38]Algorithmsnowshapehowwemove,powerourhomes,accesshealthcareandrespondtoemergencies.[12,19,20,21]Yetasdigitalsystemsconvergewithphysicaloperations,risksmultiply:cyberintrusions,adversarialAI,cascadingoutagesandgovernanceblindspotsthaterodetrustinbothsystemsandinstitutions.

AIinsmartcitiesdifferssharplyfromenterpriseorresearchenvironments.Thesesystemsrunoncontinuousdataflows,atmission-criticalspeedsandacrossinterdependentinfrastructures.Acompromisedalgorithm,whetherinwelfareeligibility,policing,housingorhealthcare,canreplicateinequitiesandtriggerfailureswithreal-worldconsequences.Relianceonproprietary,opaqueorthird-partycodefurtherlimitsvisibilityandaccountability.[1]Withoutauditing,transparencyandcitizenrecourse,citiesriskembedding“invisibleinfrastructure,”systemsthatshapedailylifewithoutpublicoversight.[2,24,26,27]

TheHISPIProjectCerebellumTAIMdirectlyaddressesthesegapsbyembeddingtamper-resistant,transparentandparticipatorygovernancemechanismsintoAIdesignanddeployment,ensuringthatresilienceisbuiltintoboththetechnologyanditsoversight.[40]

Tonavigatethesedynamics,thispaperappliestheAI-EnabledSmartCityCybersecurityFramework,outlinedinFigure1,asitsguidinglens.[7,9,10]

Figure1:AI-EnabledSmartCityCybersecurityFramework

12

Themodelidentifiessixinterdependentdimensions,perTable1.[1,3,59,10,14,25]

DESCRIPTION

Life-cycleprotectionstoensuretransparency,encryptionandaccountability.

Guardrailsforethical,transparentandlawfulAIadoption.

Resilienceagainsthybridthreatswheredigitalfailurestriggerreal-world

consequences.

Theassurancelayerconnectingcitysystemsandpartnersthrough

secureidentity,dataexchange,deviceattestation,supply-chainintegrity,

segmentationandzero-trustacrossIT,OTandcloud/edge.

24/7IT/OTfusiontomonitor,defendandrestoreservices:monitoring,intel/

hunting,incidentresponse,vuln/patch/config,securedata/AI&machine

learningoperations(MLOps),exercisesandmetrics.

MODELELEMENTS

AI+DataTrust&

Privacy

AI-EnabledGovernance&Compliance

Cyber-PhysicalInterdependence

InfrastructureTrust

Network

SmartCityCyberCoreOperations

Workforce&Community

Preparingpeopleasbothguardiansandbeneficiariesofsecuresmart

systems.

Threekeydimensionstotheframeworkinclude:

•OperationalizationMetrics:Howmaturityandresiliencearequantified

•Cross-SectorInteroperability:Howframeworksperforminmultijurisdictionalcontexts

•AssuranceandAuditability:Howtrustismaintainedthroughcontinuousvalidation

Theaimistotransformpolicyguidanceintorepeatablepracticeandtoequipcityleaderswithmethodsfortrackingprogressagainstdefinedgovernancebenchmarks.Thismodelalsoemphasizesthatresilienceisnotjusttechnical;itrequiresanticipatorygovernance,life-cycleAIriskmanagementandcross-sectorcollaboration.[10,14,18]

DrawingontheU.S.NationalCybersecurityStrategy(2023),theNISTAIRMF(2023),PresidentialPolicyDirective21onCriticalInfrastructureSecurityandResilience(2013),andglobalbestpracticesfromtheOrganizationforEconomicCooperationandDevelopment(OECD),WorldEconomicForum(WEF)andEuropeanUnion,thiswhitepaperappliestheseinsightstothe16U.S.criticalinfrastructuresectors.IthighlightsboththeopportunitiesAIoffersandthevulnerabilitiesitintroduces.[10,14,18,28,34,38]

13

Ourthesisisclear:Smartcitiescannotbetruly“smart”withoutbeingsecure,accountableandcitizencentered.[14,24,26,27]Governancemustbeanticipatory,notreactive;inclusive,notsiloed;andresilient,notbrittle.Thefutureofsmartcities,andthetrustoftheircitizens,dependsonembeddingfairness,transparencyandethicsintotheirdigitaltransformationtoday.[1,25]

CalltoAction:Urbanleaders,policymakers,technologistsandcivilsocietymustactcollectively.Thetimeforsiloedinnovationisover.Buildingresilient,trustworthyandinclusivesmartcitiesdemandsurgent,coordinatedresponsibility,andthatworkstartsnow.

“CitiesmustgovernAInotjustasatechnicalasset,butasapublicgoodthatshapesdemocraticparticipation,equityandresilience.”–OECD.AI,2024[34,36]

14

2.0EVOLVINGSMARTCITYTHREATLANDSCAPE

Smartcities,envisionedashubsofAI-enabledefficiency,nowstandatacrossroads.TheintegrationofAI,OTanddigitalinfrastructuredeliverstransformativebenefits,real-timetransitoptimization,predictivemaintenance,precisionenergyuseandautomatedemergencyresponse,butalsowidenstheattacksurfaceandheightensinterdependenciesthattraditionalcybersecurityframeworkscannotmanage.[54]

“Smartcitiesarenotjustdigital;they’repoliticalandethicalbattlegroundsforhowdata,safetyandfairnessaredistributed.”—AIEthics&UrbanPolicyBrief

Urbansystems,fromwatertreatmenttoemergencydispatch,areincreasinglyreliantondistributedAI,globalsupplychainsandsensornetworks.Asinglecompromisednode,whetherthroughdatapoisoning,sensorspoofingorunpatchedOT,canrippleacrosssectors.Thetraditionalcybersecurity“CIAtriad”(confidentiality,integrity,availability)isnolongersufficientwhentherisksextendintophysical,politicalandhumanitariandomains.[51,53-56,62,63,67,69]

Equallyimportant,thenatureoftheserisksvariesbygovernancescaleandmaturity.Metropolitansmartcitiesdeployhighlyintegratedsystems,whichmagnifyinterdependenceandtheriskofcascadingfailures.Incontrast,smallercommunities,includingvillages,incorporatedtowns,suburbanjurisdictionsandruralmunicipalities,areembracing“smart”technologiesthroughfragmented,service-by-serviceadoption.Taxcollectionplatforms,ruraltelehealthtools,schoolcommunicationsystemsandlocalcivilcourtapplicationsoftenemergeinisolation,stitchedtogetherwithlimitedfundingandstaffing.[54,67]

Thispatchworkapproachcreatessiloedinfrastructureswithfragmentedsecuritycontrols.Vulnerabilitiesinonesystem,suchasweakauthenticationinaschoolplatformorunpatchedruralhealthapplication,canbeexploitedaspivotsintobroaderregionalorevennationalnetworks.Becausesmallerjurisdictionslackthedepthofcybersecurityresourcesavailabletometropolitancenters,theirexposurebecomesaforcemultiplierforsystemicrisk.

AsRANDandOECDemphasize,smartcitiesmustbeunderstoodassociotechnicalecosystems:digitalinfrastructuresinseparablytiedtogovernancecapacity,communitytrustandequity.Withoutmaturegovernanceandresourcedoversight,bothlargeandsmalljurisdictionsalikefaceescalatingrisks.ThethreatlandscapenowencompassesnotonlyconventionalcyberattacksbutalsoadversarialAI,poisoneddata,generativedisinformation,algorithmicbias,shadowdeploymentsandthefragilityoflegacyintegration,allofwhicharemagnifiedwhengovernancecapacityisuneven.[51,54,61,69]

Theriskprofilewillintensifyashigh-riskAImodalitiesmovefrompilotstomainstreamuse,especiallyaffectivecomputingandmetaverse/XRinterfaces.Affectivesystems(empathicoremotion-sensingAI)caninfer,simulateandevenshapehumanbeliefsandbehaviors,sometimeswithoutmeaningfulconsent,raisingmaterialconcernsforfairness,privacyandmanipulation.[14,57,67]

15

Metaverse/XRtechnologies,whichblendphysicalandimmersivedigitalenvironments—engagepeopleatadeeplyhumanlevel,collapsingboundariesbetween“real”and“virtual”contextandintroducingnovelattacksurfacesacrossidentity,safetyandprovenance.[58-60]Together,thesemodalitiescreatecognitiveandsociolegalvulnerabilities:manipulationofattentionandjudgment,erosionofinformedconsent,contestednotionsofharmandambiguousaccountabilitywhendecisionsarecoproducedbyhumansandmachineagents.[14,57,61]Atscale,theywillstrainexistingsecuritycontrolsandoversight,fromcontrolbaselinesandincidentresponsetosupply-chainassurance,demandingaugmentationoftoday’sframeworksforAI,OT/ITandcommunicationssecurity.[21,22,55]

Table2(EmergingThreatModalities)and

AppendixC

expandonthesevectorsandmapthemtoconcreteriskcontrols,auditmechanismsandpolicyresponsesforurbanAIdeployments,offeringstructuredreferencestosupportriskassessments,governancestrategiesandpolicyresponses.[41,55,56]

Table2:EmergingThreatModalities

THREATMODALITYDESCRIPTIONEXAMPLE/IMPACT

AdversarialAI&ModelManipulation[10,11,54]

ImperceptibleinputsmisleadAImodels.

Autonomousvehiclesmisclassifytrafficsigns,

destabilizingnavigationandsafety.

Affective/ExtendedReality(XR)Manipulation[57-59,67]

Emotion-sensingAIandimmersive

metaverse/XRexperiencesinferand

shapeuserstates,sometimeswithout

meaningfulconsent.

Targeted“nudges”incityappsorimmersive

alertssteercrowdbehaviorduring

emergencies;immersivemisinformationblurs

reality,underminesjudgmentandpublictrust.

AlgorithmicBias[11,12,14]

Unfairorbiasedmodelsencodesystemic

inequities.

Discriminatoryoutcomesinhousingallocation,

emergencyresponseorwelfaredelivery.

DataPoisoning&SensorSpoofing[11,21]

Compromisedtrainingorreal-timedata

degradesintegrity.

Manipulatedinputsdisruptenergybalancing,

floodmonitoringorbuildingautomation.

FragmentedLocalAdoption

[51,62,63,67,69]

Siloedsystemsimplementedin

smalljurisdictionswithoutintegrated

governance.

Disconnectedruralhealth,schoolandcourtappsexposingcommonauthenticationgaps.

GenerativeAI-EnabledThreats[60,61]

Malicioususeofgenerativemodelsfor

attack.

Deepfake-drivendisinformation,adaptivemalwareorvoicecloninginfraudschemes.

IndustrialControlSystem

(ICS)/OTVulnerabilities[21,54,55]

BlindspotswhenAIinteractswithcontrol

systems.

AI-drivenanomaliesinwatertreatmentor

energydistributionevadedetection.

LegacyIntegrationFragility

[56,69,64]

Smartoverlaysonoutdatedinfrastructure

lackcoherence.

RetrofittedtrafficgridsfailunderAI-driven

optimizationattempts.

ShadowAIDeployments[10,

31,32]

Toolsdeployedoutsidegovernanceor

riskassessment.

EmergencyservicesadoptingunvettedAI

withoutredteamingorbiasaudits.

SupplyChainRisk[22,53]

Vulnerabilitieshiddeninvendor-supplied

hardware/software.

Firmwarebackdoorsembeddedincritical

infrastructurecomponents.

SurveillanceOverreach[24,27]

AI-drivenmonitoringexpandsbeyond

ethical/legalbounds.

Masscitizentrackingcreatesexploitation

risksanderodespublictrust.

16

2.1ExpandingCyber-PhysicalRisk

Urbansystems,fromwatertreatmenttoemergencydispatch,areincreasinglyreliantondistributedAI,globalsupplychainsandsensornetworks.Asinglecompromisednode,whetherthroughdatapoisoning,sensorspoofingorunpatchedOT,canrippleacrosssectors.[41]Thetraditionalcybersecurity“CIAtriad”(confidentiality,integrity,availability)isnolongersufficientwhentherisksarephysical,politicalandhumanitarian.[24-27]

Beyondthesesystem-levelrisks,thediffusionofsmart-citycapabilitiesintosmallcommunities,villages,smaller-incorporatedcitiesandsuburbanorrurallocalities,createsadistinctexposureprofile.Thesejurisdictionsoftenfaceconstrainedbudgetsandstaffing,adoptdigitaltoolsoneserviceatatime(e.g.,taxcollection,ruralhealth,schoolcommunications,civilcourts)andintegratethematopagingOT.Theresultisapatchworkof