解读欧盟人工智能法案（英）-毕马威.docx

《解读欧盟人工智能法案（英）-毕马威.docx》由会员分享，可在线阅读，更多相关《解读欧盟人工智能法案（英）-毕马威.docx（27页珍藏版）》请在优知文库上搜索。

1、umfrJrfWDecodingtheEUAlActUnderstandingtheAlAct,simpactandhowyoucanrespondKPMG.MaketheDifference.KPMGInternationalContents07ExecutivesummaryExaminingtheAlAcfsimpactandscopeUnravellingtheAlAcfskeycomponentsNextstepsIntroductionArtificial Intelligence (AI) is offering new benefits to society and busin

2、esses, aiming to transform the workplace and major industries along the way.Simply put, the race is on to embrace the remarkable and evolving power of Al and automation.AsKPMGsGlobalTeChReDOrt2023reveals,mostglobalexecutives(62percent)reportanincreaseinperformanceorprofitabilityfromdigitaltransforma

3、tioninitiativesrelatedtoAlandmachinelearningoverthepast24months.And68percentsaythesetechnologieswillplaya*vital,roleinhelpingthemachievetheirbusinessobjectivesoverthenextthreeyears,while57percentbelieveAlandmachinelearningwillbeimportant*inmeetingshort-termobjectives.ButastheworldwideAlproliferation

4、inbusinessandoreverydaylivesunfolds,thereisacriticalneedforguardrailsandlegislationtodealwithsignificantnewrisksregardingtheappropriateandethicaluse,developmentanddistributionofAl.AccordingtoTnJStinartificialintellience,aglobalsurveyconductedbyKPMGAustraliaandtheUniversityofQueensland,threeinfivepeo

5、plearewaryabouttrustingAlsystems,and71percentexpectAltoberegulated.Morerecently,CEOsfromglobaltechgiantscalledforgreaterAlregulationatameetingonCapitolHilltoprotectpeoplefromtheworsteffectsofAl.Inresponse,theEuropeanUnion(EU)hasreachedaground-breakingprovisionalagreementonacomprehensiveArtificialInt

6、elligenceAct(AlAct)thattakesarisk-basedapproachtoprotectingfundamentalrights,democracy,theruleoflawandenvironmentalsustainability.iThoughitsexpectedtobecomelawin2024,withcomplianceexpectedby2025,thislegislationthefirstofitskindisanticipatedtoemergeasthede-factonewglobalstandardforAlregulation.Withth

7、eintroductionoftheAlAct,theEUaimstostrikeabalancebetweenfosteringAladoptionandensuringindividualsrighttoresponsible,ethicalandtrustworthyuseofAl.Inthispaper,weexplorewhattheAlActmaymeantoyourorganizationandexaminethestructureoftheAlAct,theobligationsitimposes,thetimelinesforcomplianceandtheactionpla

8、nthatorganizationsshouldconsider.OrganizationalleadershipshoulddriveinitiativesinlinewiththeAlAct,companybrand,valuesandrisktolerancetopromoteresponsibleuseofAl.Thiscanhelppromoteethicaldevelopment,regulatorycompliance,riskmitigationandstakeholdertrust.，DavidRowlandsGlobalAlLeaderKPMGInternationalAl

9、shouldbedevelopedandusedwithafocusonsafetyandethics,turningtechnologicaladvancementintoapositiveforceforsociety.TheEUAlActwillhelpfosterinnovationwhileprotectingend-users.，1.aurentGobbiGlobalTrustedAlLeaderKPMGInternationaliEuropeanParliament.(December9,2023).ArtificialIntelligenceAct:dealoncomprehe

10、nsiverulesfortrustworthyAlPressrelease.ExecutivesummaryTheAlActaimstoregulatetheethicaluseofAlAlholdsimmensepromisetoexpandthehorizonofwhatisachievableandtoimpacttheworldforourbenefit-butmanagingAsrisksandpotentialknownandunknownnegativeconsequenceswillbecritical.TheAlActissettobefinalizedin2024anda

11、imstoensurethatAlsystemsaresafe,respectfundamentalrights,fosterAlinvestment,improvegovernance,andencourageaharmonizedsingleEUmarketforAl.MostAlsystemsneedtocomplywiththeAlActbythefirsthalfof2026TheAlActsdefinitionofAlisanticipatedtobebroadandincludevarioustechnologiesandsystems.Asaresult,organizatio

12、nsarelikelytobesignificantlyimpactedbytheAlAct.Mostoftheobligationsareexpectedtotakeeffectinearly2026.However,prohibitedAlsystemswillhavetobephasedoutsixmonthsaftertheAlActcomesintoforce.Therulesforgoverninggeneral-purposeAlareexpectedtoapplyinearly2025.2Providersandusersofhigh-riskAlsystemsfacestri

13、ngentobligationsTheAlActappliesarisk-basedapproach,dividingAlsystemsintodifferentrisklevels:unacceptable,high,limitedandminimalrisk.3High-riskAlsystemsarepermittedbutsubjecttothemoststringentobligations.Theseobligationswillaffectnotonlyusersbutalsoso-calledprovidersofAlsystems.Theterm,provider,inthe

14、AlActcoversdevelopingbodiesofAlsystems,includingorganizationsthatdevelopAlsystemsforstrictlyinternaluse.Itisimportanttoknowthatanorganizationcanbebothauserandaprovider.Providerswilllikelyneedtoensurecompliancewithstrictstandardsconcerningriskmanagement,dataquality,transparency,humanoversight,androbu

15、stness.UsersareresponsibleforoperatingtheseAlsystemswithintheAlActslegalboundariesandaccordingtotheprovidersspecificinstructions.Thisincludesobligationsontheintendedpurposeandusecases,datahandling,humanoversightandmonitoring.GuardrailsforgeneralAlsystemsNewprovisionshavebeenaddedtoaddresstherecentad

16、vancementsingeneral-purposeAl(GPAI)models,includinglargegenerativeAlmodels.”ThesemodelscanbeusedforavarietyoftasksandcanbeintegratedintoalargenumberofAlsystems,includinghigh-risksystems,andareincreasinglybecomingthebasisformanyAlsystemsintheEU.ToaccountforthewiderangeoftasksAlsystemscanaccomplishandtherapidexpans