版權(quán)說明:本文檔由用戶提供并上傳,收益歸屬內(nèi)容提供方,若內(nèi)容存在侵權(quán),請進(jìn)行舉報(bào)或認(rèn)領(lǐng)
文檔簡介
ThePromiseandPeriloftheAIRevolution:ManagingRisk
?2023ISACA.AllRightsReserved.
Risk
2THEPROMISEANDPERILOFTHEAIREVOLUTION:MANAGINGRISK
CONTENTS
4
WhenCorporateWisdomMeetsArtificialIntelligence
4
TheCurrentStateofArtificialIntelligence
6
ToAIorNottoAI
6
IdentifyingAIRiskandReward
7
ConductinganAIBenefitAnalysis
8
IdentifyingAIRisk
9/SocietalRisk
10/IPLeakageandInvalidation10/InvalidOwnership
11/CybersecurityandResiliencyImpact11/WeakInternalPermissionStructures11/SkillGaps
12/Overreactions
12/IntendedandUnintendedUse12/DataIntegrity
13/Liability
14
AdoptingaContinuousRiskManagementApproach
14/StepOne:IdentifyRisk
14/StepTwo:DefineRiskAppetite
15/StepThree:MonitorandManageRisk
15
BuildingAISecurityPrograms:EightProtocolsandPractices
15/One:TrustbutVerify
16/Two:DesignAcceptableUsePolicies16/Three:DesignateanAILead
16/Four:PerformaCostAnalysis
17/Five:AdaptandCreateCybersecurityPrograms
17/Six:MandateAuditsandTraceability18/Seven:DevelopaSetofAIEthics
18/Eight:SocietalAdaptation
19
ProsperinginanAI-PoweredFuture
21
Acknowledgments
?2023ISACA.AllRightsReserved.
3THEPROMISEANDPERILOFTHEAIREVOLUTION:MANAGINGRISK
ABSTRACT
Itisthepremiseofsomanysciencefictionmovies:technologysurpasseshumanintelligence,wreakshavocand
ultimatelytakesoverhumankind.Whilethatmoviehasnotyethappenedinreallife,therecentreleaseofChatGPT
feltliketheopeningcredits–andthatisjustoneofmanygenerativeartificialintelligence(AI)tools.Nowhigh-profile
techleaderssuchasAppleco-founderSteveWozniakandTeslachiefexecutiveofficer(CEO)ElonMuskareamongthoseaskingcompaniestoholdbackon“giantAIexperiments”1whiletheindustryconductsariskassessment.Theiropenletterstatesthatlarge-scaleAIprojects“canposeprofoundriskstosocietyandhumanity”withoutoversightandintelligentmanagementandaskedforatemporarypauseonfurtherdevelopment.TheletteralsostressedtheneedforgovernancesystemsandanewregulatoryauthoritydedicatedtoAI.
Still,otherindustryleadersbelieveboththebenefitsandrisksofgenerativeAIhavebeenexaggerated2–thatwearedealingwithlimitedtechnologiesthataren’tnearlyashelpfulasourbiggesthopesoraspowerfulasourworstfears.Butoneconclusioniscertain:AIisalreadysweepingthroughourbusinessesandourworld,andtheneedforchiefinformationsecurityofficers(CISOs),ITriskmanagers,executivesandITseniormanagementtokeeppacewiththerapidlyevolvingrisklandscapeisurgent.
1F,“PauseGiantAIExperiments:AnOpenLetter,”22March2023,
/open-letter/pause-giant-ai-experiments/
2Rundle,J.;“CybersecurityChiefsNavigateAIRisksandPotentialRewards,”TheWallStreetJournal,25May2023,
/articles/
cybersecurity-chiefs-navigate-ai-risks-and-potential-rewards-9138b76d
?2023ISACA.AllRightsReserved.
4THEPROMISEANDPERILOFTHEAIREVOLUTION:MANAGINGRISK
WhenCorporateWisdomMeetsArtificialIntelligence
Justmonthsago,itwashardtoimaginethatprofession-alsfromeveryindustrywouldturntooneapplicationtogeneratelegalargumentsandadvice,computercode,
Shakespearean-stylesonnetsandclinicaltreatment
plans.YetmillionsofusershavequicklyacclimatedtotheconvenientnewworldofgenerativeAItoolssuchasOpenAI’sChatGPT3andGoogle’sBard.4Despiteitsim-mensesuccessandmyriadofuses,mostusersarenotfamiliarwiththedangersofgenerativeAI.Consequently,ascompaniesleapintoawhite-hotracetodesignandutilizeAItools,manyriskmanagementeffortsarefallingbehind.
Theefforttoestablishanduseemergingtechnology
shouldnotcomeattheexpenseofriskmanagement.Unfortunately,thisisapatterncommontomostnewtechnologies.Thosewhoremembertheadventof
smartphones,3DprintingortheInternetofThings(IoT)
mayassumetheemergenceofAIisonparwiththosebreakthroughs.ButitismoreakintothecreationoftheautomobileortheInternet:ablazeofinnovationthatwillengulfeveryindustry.
Theefforttoestablishanduseemergingtechnologyshouldnotcomeattheexpenseofriskmanagement.
AIushersinafoundationalshiftinhowweengagewithtechnology.Thetraditionalguardrailsandsecurityproto-colsthathaveservedusuptothispointarenowtoacer-tainextentinadequate,asgenerativeAIhasonlygrowninpopularityandinfluence.Newterritoryincludesnew
risk,andunlikepreviousquantumleapsintechnology,businesseslookingforavoiceofauthorityareinsteadfindinguncertaintyabouthowtothoughtfullymaximizeAIvaluewhiledeftlyminimizingrisk.
TheCurrentStateofArtificialIntelligence
DespiteAI’sabilitytomimichumanthoughtpatternsandspeech,noneofitissentient.Mostofwhatiscommon-lyreferredtoasAIisactuallycomprisedofmachine
learningtechniquesorlargelanguagemodels(LLMs).
Forexample,afterbeingfedadatasetof300billion
words,5ChatGPTconsumedbooks,websites,WikipediaandothersourcesandwastrainedusingReinforcementLearningfromHumanFeedback(RLHF),withthefeed-backimprovingitsresponses.BardwastrainedinsimilarwaysbutalsodrawsitsinformationfromtheInternet.
Megatron,6ajointventurebetweenMicrosoftandNvidia,
wastrainedthroughnovel“parallelism”techniquesandcouldsoonsurpassChatGPTthrougharadicalincreaseincapabilities.Thetechniquesusedtodevelopthese
toolsraisequestionsaboutwherethedatausedinLLMsiscomingfromandwhethertotrustthisdatatobeusedtotrainAI-enabledtools.
WhileaprivatecompanylikeChatGPThasclaimedthelion’sshareofthemediaspotlight,startupslike
ChatSonic,Jasper,WordtuneandothersarehelpingusersthinkofAInotasonetoolbutasanewtypeof
3O,“IntroducingChatGPT,”30November2022,
/blog/chatgpt
4,“TryBard,anAIexperimentbyGoogle,”
/
5Iyer,A.;“BehindChatGPT’sWisdom:300BnWords,570GBData,”AnalyticsIndiaMagazine,15December2022,
/
behind-chatgpts-wisdom-300-bn-words-570-gb-data/
6D,“Megatron-TuringNaturalLanguageGeneration,”NVIDIADeveloper,3October2022,
/mega
-tron-turing-natural-language-generation
?2023ISACA.AllRightsReserved.
5THEPROMISEANDPERILOFTHEAIREVOLUTION:MANAGINGRISK
functionaltechnology.ThisisinpartthroughgenerativeAIpre-trainingthatusesinformationcontributionsfromtheprivateandpublicsectors.Thesecompaniesare
revolutionizingthewayweinteractwithAIbyprovidingarangeoftoolsandservicesthatcatertothediverseneedsofusers.Inresponsetouserfeedback,creatorsareaddingsecuritycontrols,textclassifiersandotherfeaturestobecomemorecompetitiveanduseful.
Forinstance,Bardisdesignedtoalsoworkasapersonalassistantandcancompletetasksrangingfrombook-
ingaholidaytocreatingamealplan.ChatGPTisbeingintegratedintoMicrosoftOffice365asMicrosoft365Copilot.7Soon,anyoneusingMicrosoftWord,ExcelorPowerPointwillbeabletouseCopilottocreateanewpresentation,developgo-to-marketstrategiesordistilleventhemostcomplexfinancialdataintoareport.
ThespeedofAIevolutionhasdazzledeventhemost
seasonedoftechveterans.NearlyeveryweekbringsanewpronouncementonAI’sadvantages,dangers,limita-tionsorpotential,withoftencontradictoryconclusions.ItisnotasurprisethatmanybusinessleadershaveoptedtowaitfortheAIdusttosettlebeforedesigningaformalbusinessstrategy.Butwhilethismayseemlikethesaf-estpath,adelaycarriesariskofitsown.
Forthisreason,nomatterwhatshapetheAIrevolutiontakes,weknowthatAIisheretostayandthatthesafestandsmartestpathtosecurityistobeginadaptingnow.BusinessleadersshouldassumethatAIisalreadybeingusedwithintheenterprise,asitspopularityhasonly
grownexponentiallywithinthepastcoupleofyears.Anexampleoftheexplosivegrowth8intheuseofChatGPTisshowninfigure1.
FIGURE1:AIUseBreakstheSpeedofSound
November2022
ChatGPTreleased◆
forpublicuse
◆
1.8billionvisits
Day5afterlaunch
Morethan1millionusers
…December2022
April2023
266millionvisits
◆June2023
100millionactiveusers
◆
7Spataro,J.;“IntroducingMicrosoft365Copilot–yourcopilotforwork,”TheOfficialMicrosoftBlog,16March2023,
/
blog/2023/03/16/introducing-microsoft-365-copilot-your-copilot-for-work/
8Ruby,D.;“30+DetailedChatGPTStatistics–Users&Facts,”DemandSage,7July2023,
/chatgpt-statistics/
?2023ISACA.AllRightsReserved.
6THEPROMISEANDPERILOFTHEAIREVOLUTION:MANAGINGRISK
ToAIorNottoAI
Thehorseisheretostay,buttheautomobileisonlyanovelty—afad.9
—AdvicefromMichiganSavingsBanktoHenryFord’slawyerHoraceRackham
Someorganizations,scaredoftheunknown,havetakentoignoringAI’sgrowinginfluence.Others,afraidof
anothertechnologicalgimmick,dismissAIpromisesasmerehype.Butwieldedwell,AItoolscanaugmenthu-manideas,complementbusinessprocessesanddistillcomplexinformationintodigestiblecontent,allofwhichcandrivehigherprofitandperformance.
Fromthewheeltotheautomobile,historyshowsthatanytechnologythatletsusdomorewithlesseffort
hasbeenwildlypopular–asgenerativeAIhasproven.Fromautomatingmundanetaskstogeneratingcreativecontent,generativeAIhasdemonstrateditspotential
toenhanceproductivityandunleashhumancreativi-
ty.Already,someworkersviewusingAIasnormalascheckingemailorpurchasinggoodsonline.ItisonlyamatteroftimebeforeAIisenmeshedwithourpersonalandprofessionallives.
Despiteitsinfiltrationintoalmosteveryfacetofsocietyandculture,somebusinesseshavebannedorrestrictedAIuse.StackOverflow,whichoffersaforumforcodingquestionsandanswers,temporarilybannedChatGPT10becauseofitstendencytogiveincorrectanswersthat
soundconfidentandlegitimate,leadinguserstobe
confusedormisleadwhenlookingforrelevantanswerstotheirquestions.SamsungElectronics,11Apple,12
JPMorganChase13andVerizonCommunications14haveheavilyrestrictedtheworkplaceuseofgenerativeAIoversecurityfears.
Whiletheserestrictionsmakesensegiventhecurrent
ambiguityofAIcapabilitiesandconsequences,most
businessleadersrealizedecisionsregardingthelong-termAIlandscapewillbefarmorecomplexthanasimpleyesorno.IfanenterprisebansAIcompletely,itcouldminimally
riskitscompetitiveadvantageassomeemployees,partners,distributorsandcompetitorsmaybenefitfromit.Evenif
AI-enabledtoolsarenotusedintheenterprise,ignoringthistechnologycanleaveanenterprisemorevulnerabletorisk(e.g.,obsolescence,loweraccesstotoptalent,etc.).
Fromthewheeltotheautomobile,historyshowsthatanytechnologythatletsusdomorewithlessefforthasbeenwildlypopular–asgenerative
AIhasproven.
Forthisreason,seniorleaders,iftheywishtoutilizeAItechnologies,willneedtoensuretherightinfrastructureandgovernanceprocessesareinplaceattheir
organizations.ToaccuratelyunderstandthevulnerabilitiesandadvantagesAIcarries,leadersmustconducta
thoroughriskimpactanalysisthataccountsforthecurrentuncertaintyoftheAIlandscapeanditsfuturepower.
IdentifyingAIRiskandReward
AIispoisedtopermeatebusinessesineveryindustry.Fromsupplyingstudentswithready-madeacademicessaystowritingscriptsforHollywoodstudiosand
analyzingtheflawsinfactoryfloordesigns,AIoffers
benefitstonearlyeveryteam,fieldandorganization.Onacollectiveindustrylevel(andindividualbusinesslevel),
9Bushnell,S.T.;TheTruthAboutHenryFord,TheReilly&LeeCompany,USA,1992
10,“Temporarypolicy:ChatGPTisbanned,”
/questions/421831/temporary-poli
-cy-chatgpt-is-banned
11Cawley,C.;“SamsungRestrictsGenerativeAIUseAfterCodeLeak,”Tech.co,8May2023,
https://tech.co/news/samsung-restricts-genera
-tive-ai-use#:~:text=New%20Policy%20Bans%20Samsung%20Employeesz
12Tilley,A.;M.Kruppa;“AppleRestrictsEmployeeUseofChatGPT,JoiningOtherCompaniesWaryofLeaks,”TheWallStreetJournal,18May2023,
/articles/apple-restricts-use-of-chatgpt-joining-other-companies-wary-of-leaks-d44d7d34?mod=article_inline
13Lukpat,A.;“JPMorganRestrictsEmployeesFromUsingChatGPT,”TheWallStreetJournal,22February2023,
/articles/jpmor
-gan-restricts-employees-from-using-chatgpt-2da5dc34?mod=article_inline
14Moneylife.in,“JPMorganChaserestrictsworkersfromusingChatGPT,”23February2023,
https://www.moneylife.in/article/jpmorgan-chase-re
-stricts-workers-from-using-chatgpt/69948.html
?2023ISACA.AllRightsReserved.
7THEPROMISEANDPERILOFTHEAIREVOLUTION:MANAGINGRISK
leadersmusttakefourimportantstepstomaximizeAIvaluewhileinstallingappropriateandeffectiveguardrails:
1.IdentifyAIbenefits.
2.IdentifyAIrisk.
3.Adoptacontinuousriskmanagementapproach.
4.ImplementappropriateAIsecurityprotocols.
Ifbusinessleadersareabletofollowthesesteps,theyaremuchmorelikelytostrikeagoodbalanceofriskversusrewardasAI-enabledtoolsandprocessesareleveragedintheirenterprises.
ConductinganAIBenefitAnalysis
LeadersandphilanthropistslikeBillGatesforeseea
cornucopiaofAIbenefits,15frombringinghealthcareandeducationtotheunderservedtorectifyingtheimpactsofclimatechange.Forenterprises,potentialadvantagesincludebetterinnovation,efficiency,productivityand
workforceoptimization.CompaniesarealreadylookingtoautomatetaskswithAIandreallocatestafftomorevalue-drivingworkwheretheycan,forexample,focusondesigningaremarkablenewservicelineordevelopingmoreattentiveclientrelationships.
CompaniesthatwishtooptoutofleveragingAI
willmostlikelyfindthemselvesonapathtoobsoles-cence.JustasAIprogramscaptivatedbillionsofusersinmereweeks,thespeedatwhichtheseprogramswillchangemarketsandproductcapabilitiescouldbe
staggering.
EnterprisetitansmayfindthattheirAI-enabled
competitionissmarter,fasterandmorelethalthaneverbefore.Figure2containssomesampleusecases.16
FIGURE2:AIInsecurityComplex—IsAISmarterThanWeAre?
CHATGPTHAS:
PassedWharton’s
MBAexamProvided
complexclinicaltreatmentplans
Produced
creativewriting
work—fromentrance
examessaystosoftwarecode
Passedthe
USMedical
LicensingExam
Passedlawschoolexams
15Gates,B.;“TheAgeofAIhasbegun,”GatesNotes,21March2023,
/The-Age-of-AI-Has-Begun
16Ault,A.;“AIBotChatGPTPassesUSMedicalLicensingExamsWithoutCramming–UnlikeStudents,”Medscape,26January2023,
/viewarticle/987549;
Sloan,K.;“ChatGPTpasseslawschoolexamsdespite‘mediocre’performance,”Reuters,
25January2023,
/legal/transactional/chatgpt-passes-law-school-exams-despite-mediocre-performance-2023-01-25/;
businesstoday.in,“‘NextstepforAI’:ChatGPTclearsUSMedicalLicensingExamandWharton’sMBAexam,”BusinessToday,25January2023,
https://www.businesstoday.in/technology/story/next-step-for-ai-chatgpt-clears-us-medical-licensing-exam-and-whartons-mba-exam-367569-2023-01-25;,“AIisPassingMedSchoolExams,”24January2023,
/chatgpt/
;Whitford,E.;“Here’sHow
ForbesGotTheChatGPTAIToWrite2CollegeEssaysIn20Minutes,”Forbes,9December2022,
/sites/emmawhitford/
2022/12/09/heres-how-forbes-got-the-chatgpt-ai-to-write-2-college-essays-in-20-minutes/?sh=313d2f3b56ad;Hutson,M.;“AIlearnstowrite
computercodein‘stunning’advance,”Science,8December2022,
/content/article/ai-learns-write-computer-code
-stunning-advance
?2023ISACA.AllRightsReserved.
8THEPROMISEANDPERILOFTHEAIREVOLUTION:MANAGINGRISK
UseofAIToughensCompetition
CompanyAisanindustryleaderthathasdominateditsmarketfor10yearsonthestrengthofoneflagshipproduct.TheCEOfeelssecureinthecompany’s
position,ascompetitiveintelligenceassuresher
thatnoothercompany’sresearchanddevelopmentapproachesthefunctionalityofCompanyA’sproduct.
ThenanupstartnamedCompanyBusesAItodevelopresearch,designagroundbreakingproduct,create
pricingmodelsandmarketingplans,launchanew
websiteandpopulateonlinesaleschannels–allwithintwoweeks,withminimalworkforce.
Withinjusttwostaffmeetings,CompanyA’sproductbecomesviewedasoutdated,itscustomerbase
decampsforCompanyBandprofitsbegintoplunge.OtherupstartsbegintouseAItodethroneCompanyB.Themarketshiftsradicallywithinafewmonths,withCompanyAfightingandfailingtoremainrelevant.
Toconductanexhaustivebenefitanalysisof
anytechnology,leadersshouldlookbeyond
simplisticadvantagesandrevisitthefoundational
ethosoftheircompanies.Someconsiderationsinclude:
?Whatdoesthecompanydo,andwhydotheydoit?
?Howdotheydoit?
?Howarethecompany’scompetitorsleveragingAItools?
?Whattoolsandtalentsdotheyrelyon,andwhataretheirmetricsforsuccess?
?HowmuchwillinvestmentinAIcost,andhowmuchreturnoninvestmentdoesitdeliver?
?Howconfidentisthecompanythatitssecurityprotocols
andmechanismswillhelpensuredataquality,integrityandconfidentiality(e.g.,intellectualproperty,personalidentifiable
information,etc.)?
?Howmightthecompany’scurrentstrategyandobjectivesevolvewhenAIremovessignificantcostandtalentbarriers?
?WhatcouldstaffandpartnersbringtotheAIinnovationtable?
TheanswerstothesequestionswillprovidearoadmapthatconnectsAIusecasestothecompany’smissionandgoals.
IdentifyingAIRisk
ThepotentialriskassociatedwithAIhasbeenpredom-inantlycharacterizedasafutureconcern.However,
proactivelyaddressingthisriskcanhelptoensurethesafeandresponsibledevelopmentofAItechnologyforuseinthebusiness.
Dr.GeoffreyHinton,commonlyreferredtoasthe“godfa-ther”ofAI,recentlyresignedfromGooglewithaformaladmissionofregretabouthiswork,callingsomeofthe
dangers“quitescary.”17Hisworkonneuralnetworksanddeeplearningprovidedthefoundationforwhatwenow
callAIprogramming;hehasoutlinedwaysfutureAIcapa-bilitiescouldexceedourunderstandingandthereforelimitourabilitytomobilizeorinstituteeffectivesafeguards.
Despitetheimportanceofrecognizingfuturerisk,thepresentstateofAItechnologyandourutilizationpat-ternspresentimmediateriskthatmustbeaddressed.
17Vallance,B.;C.Vallance;“AI‘godfather’GeoffreyHintonwarnsofdangersashequitsGoogle,”BBCNews,
/news/
world-us-canada-65452940
?2023ISACA.AllRightsReserved.
9THEPROMISEANDPERILOFTHEAIREVOLUTION:MANAGINGRISK
Thisriskcanbebroadlysummedup:AIcanexacerbateanyexistingissues,suchasalackofqualitycontrolor
poordataintegrity,andsystemswillbevulnerableto
cyberattacksandmayintroducenewones.Fromdata
breachestoCEOfraud,AIcanquicklyamplifya“con-
trolled”risktoachaoticlevel,potentiallyderailingan
unpreparedbusiness.SomenotableexamplesofhowAIischangingbusinessfunctionscanbefoundinfigure3.18
HoweverusefulAImightbe,itisnotwithout
inherentrisk.KeyaspectsofAIriskarecoveredinthefollowingsections.
SocietalRisk
AsAIcontinuestoinfluencesociety,wemustremain
criticalofitsaimsandpotentialharm.AIcanbeused
togeneratedisinformationandmanipulatepopulationsandpoliticaldemographicsinaconvincinganddamag-ingway.AIhasalreadybeenusedintheexecutionof
newtypesofcybercrime(e.g.,deepfakes,sophisticatedphishingemails,AIvoicegenerators,etc.).Forexample,afakeimageofanexplosionatthePentagonwaswidelycirculatedonTwitter,causingthestockmarketto
react.19ThisincidentillustrateshowAI-generated
FIGURE3:Disruption,DisplacementandDiscovery—HowMuchWillAIChangeBusiness?
GoldmanSachsestimates
AIcanreplace300million
jobsor28%ofthe
USworkforce
GenerativeAI
Studiesshow:
>SoftwareengineerscancodetwiceasfastusingAItools
>Economistscanbe
10–20%moreproductiveusingLargeLanguageModels
>Callcenteroperatorswere14–30%moreproductive
isthefirst
technology
expectedto
succeed
increativity
andhuman
interactions
GenerativeAI
couldraisethe
globalGDPby7%
18Jones,J.;“AIcouldautomate300millionjobs.Here’swhicharemost(andleast)atrisk,”ZDNET,27May2023,
/article/
ai-could-automate-25-of-all-jobs-heres-which-are-most-and-least-at-risk/;Elder,B.;“SurrenderyourdeskjobtotheAIproductivitymiracle,says
GoldmanSachs,”FinancialTimes,27March2023,
/content/50b15701-855a-4788-9a4b-5a0a9ee10561;
Neil-Baily,M.;E.
Brynjolfsson;A.Korinek;“Machinesofmind:ThecaseforanAI-poweredproductivityboom,”Brookings,10May2023,https://www.brookings.
edu/research/machines-of-mind-the-case-for-an-ai-powered-productivity-boom/;Kalliamvakou,E.;“Research:quantifyingGitHubCopilot’simpactondeveloperproductivityandhappiness,”TheGitHubBlog,7September2022,
https://github.blog/2022-09-07-research-quantifying-github-copi
-lots-impact-on-developer-productivity-and-happiness/;Korinek,A.;“LanguageModelsandCognitiveAutomationforEconomicResearch,”NationalBureauofEconomicResearch,
/10.3386/w30957;
Brynjolfsson,E.;D.Li;L.Raymond;“GenerativeAIatWork,”NationalBureauof
EconomicResearch,
/10.3386/w31161;
Chui,M.;R.Roberts;L.Yee;“GenerativeAIishere:HowtoolslikeChatGPTcouldchangeyourbusiness,”McKinsey,20December2022,
/capabilities/quantumblack/our-insights/generative-ai-is-here-how
-tools-like-chatgpt-could-change-your-business
19Bond,S.;“FakeviralimagesofanexplosionatthePentagonwereprobablycreatedbyAI,”NPR,22May2023,https://www.npr.org/2023/05/22/1177590231/fake-viral-images-of-an-explosion-at-the-pentagon-were-probably-created-by-ai
?2023ISACA.AllRightsReserved.
10THEPROMISEANDPERILOFTHEAIREVOLUTION:MANAGINGRISK
misinformationcanhavefar-reachingconsequences,un-derminingthestabilityoffinancialmarketsanderodingtrustininstitutions.Itisforthisreasonthatmechanismswillneedtobedevelopedforcompaniesandusersto
beabletoweedoutmisinformation.ItisalsouptotheusersofAItoolstoevaluatetheinformationpresentedtotheminordertomitigateitspotentialnegativeimpacts.Furthermore,thepotentialforeconomicupheavalisdi-rectlylinkedtojobdisplacement;itispredictedthatover
300millionjobs20willbeimpactedbyChatGPT
models,withunemploymentaffectingmanyindustriesandprofessions.
IPLeakageandInvalidation
ManyAIusershavethoughtlesslyfedintellectualproper-ty(IP),tradesecrets,competitorcontentandotherdataintoAImodels,whichintroducesarangeofprivacyrisk.Recently,forinstance,Samsungemployeesacciden-
tallyleakedproprietarycompanyinformationbyusingChatGPT.21Anotherexamplecouldinvolveapricingandcommoditiescompanythatwantstokeepitswholesalepricesconfidential.
However,iftheyallowemployeestoutilizeAItools,
thatconfidentialinformationcouldbecomepublicifanemployeecopiesandpastesitintoanAIprogram.Addi-tionally,theaforementionedMicrosoftCopilotwillscourinternalcompanydataonSharePoint,Excelspread-
sheets,Slack,emails,TeamsmessagesandOneDrive,potentiallysharingprivateandsensitivedata.
Forthisreason,somecompanieslikeAmazonhave
warnedemployeesnottosharesensitiveinformationwithChatGPT.22Still,manymorecompanieshavenosuchpoliciesorwarnings,whichshouldincludeaclear
definitionofwhatthecompanydeemssensitivein-
formation.Thepoliciesshouldalsoensurealignment
withemployeeswhomayotherwisebeoblivioustothecopyrights,trademarksandsensitiveinformationtheyaresharingwithanenormousdigitalrepositorythatcanshareitwithothers.
Itshouldalsobementionedthatdisgruntledemployeeswhorealizetheimplicitharmsofsharingthisinforma-tionmaypurposefullyfeedtheAItooldatatopotentiallyharmthecompany’sreputation.Employeesinserting
informationintoAIapplicationscouldinvalidatepatentsandotherIPasIPprotectioniscontrolledbycomplexlaws.Forinstance,ifacompanyfailstoprotectitstradesecretsadequatelybecauseofemployeenegligence,itmayforfeitthelegalprotectionsaffordedtotheconfi-dentialinformation.
InvalidOwnership
AnycompanyusingAItocreateoutput—suchasamar-ketingtagline—mustguaranteeitistrulyoriginalIPandnotaderivativeofanotherperson’swork.23Ifnot,thesebusinessesmaydiscovertheirownershipofthenew
assetsisinvalid.Further,theUSCopyrightOfficehas
rejectedcopyrightsforsomeAI-generatedimages,24de-pendingonwhethertheywerecreatedfromtextpromptsoriftheyreflectthecreator’s“ownmentalconception.”
Invalidownershipcouldalsooccurif,forexample,awritercopyandpastesalinktoanadmirablepieceofcompetitorcontentasinspirationintoanAItool
andisgivenbackhalf-plagiarizedcontent.25Invalid
ownershipclaimsregardingAIcanultimatelyleadtolegaldisputesandadverselyimpactanenterprise’sreputation.
20OpcitJones
21Maddison,L.;“SamsungworkersmadeamajorerrorbyusingChatGPT,”Techradar,4April2023,
/news/samsung-work
-ers-leaked-company-secrets-by-using-chatgpt
22Sengupta,A.;“AmazonwarnsemployeesaboutChatGPT,saysdonotsharesensitiveinfowithchatbot,”IndiaToday,27January2023,https://www.indiatoday.in/technology/news/story/amazon-warns-employees-chatgpt-do-not-share-sensitive-info-with-chatbot-2327014-2023-01-27
23McKendrick,J.;“WhoUltima
溫馨提示
- 1. 本站所有資源如無特殊說明,都需要本地電腦安裝OFFICE2007和PDF閱讀器。圖紙軟件為CAD,CAXA,PROE,UG,SolidWorks等.壓縮文件請下載最新的WinRAR軟件解壓。
- 2. 本站的文檔不包含任何第三方提供的附件圖紙等,如果需要附件,請聯(lián)系上傳者。文件的所有權(quán)益歸上傳用戶所有。
- 3. 本站RAR壓縮包中若帶圖紙,網(wǎng)頁內(nèi)容里面會有圖紙預(yù)覽,若沒有圖紙預(yù)覽就沒有圖紙。
- 4. 未經(jīng)權(quán)益所有人同意不得將文件中的內(nèi)容挪作商業(yè)或盈利用途。
- 5. 人人文庫網(wǎng)僅提供信息存儲空間,僅對用戶上傳內(nèi)容的表現(xiàn)方式做保護(hù)處理,對用戶上傳分享的文檔內(nèi)容本身不做任何修改或編輯,并不能對任何下載內(nèi)容負(fù)責(zé)。
- 6. 下載文件中如有侵權(quán)或不適當(dāng)內(nèi)容,請與我們聯(lián)系,我們立即糾正。
- 7. 本站不保證下載資源的準(zhǔn)確性、安全性和完整性, 同時(shí)也不承擔(dān)用戶因使用這些下載資源對自己和他人造成任何形式的傷害或損失。
最新文檔
- 2024-2030年中國城市公共汽車客運(yùn)行業(yè)發(fā)展前景及投資規(guī)劃研究報(bào)告版
- 2024-2030年中國地毯行業(yè)競爭格局及未來投資趨勢分析報(bào)告
- 2024-2030年中國國際貨代行業(yè)未來發(fā)展趨勢及投資風(fēng)險(xiǎn)分析報(bào)告
- 2024年度物聯(lián)網(wǎng)(IoT)設(shè)備控制系統(tǒng)軟件開發(fā)合同技術(shù)集成與擴(kuò)展2篇
- 茂名職業(yè)技術(shù)學(xué)院《國學(xué)》2023-2024學(xué)年第一學(xué)期期末試卷
- 2024年深圳子女撫養(yǎng)權(quán)協(xié)議書樣本3篇
- 中國日報(bào)2019年9月25日
- 馬鞍山職業(yè)技術(shù)學(xué)院《美術(shù)基礎(chǔ)與欣賞》2023-2024學(xué)年第一學(xué)期期末試卷
- 呂梁學(xué)院《信息安全綜合》2023-2024學(xué)年第一學(xué)期期末試卷
- 2024年度醫(yī)院護(hù)工工作環(huán)境與職業(yè)健康保護(hù)協(xié)議下載3篇
- 分子流行病學(xué)進(jìn)展與其在疾病預(yù)防控制中的實(shí)際應(yīng)用
- 甘井子名稱的來歷大連史話(八)
- 7混凝土護(hù)坡施工方案
- 1、每日桶裝水送水登記表
- 精益生產(chǎn)現(xiàn)場管理和改善課件(共132頁).ppt
- 鋼軌閃光焊 技術(shù) 焊縫 傷損_圖文
- 風(fēng)管風(fēng)機(jī)吊架重量表
- 房屋漏水賠償協(xié)議書甲方
- 預(yù)焙陽極配料控制
- 二階低通濾波器的設(shè)計(jì)-模電課設(shè)報(bào)告
- 六類非屏蔽網(wǎng)線模塊安裝
評論
0/150
提交評論