CSA大型语言模型（LLM）威胁分类（英）-30页

LargeLanguageModel(LLM)ThreatsTaxonomyThepermanentandofficiallocationfortheAIControlsFrameworkWorkingGroupishttps://cloudsecurityalliance.org/research/working-groups/ai-controls©2024CloudSecurityAlliance–AllRightsReserved.Youmaydownload,store,displayonyourcomputer,view,print,andlinktotheCloudSecurityAllianceathttps://cloudsecurityalliance.orgsubjecttothefollowing:(a)thedraftmaybeusedsolelyforyourpersonal,informational,noncommercialuse;(b)thedraftmaynotbemodifiedoralteredinanyway;(c)thedraftmaynotberedistributed;and(d)thetrademark,copyrightorothernoticesmaynotberemoved.YoumayquoteportionsofthedraftaspermittedbytheFairUseprovisionsoftheUnitedStatesCopyrightAct,providedthatyouattributetheportionstotheCloudSecurityAlliance.©Copyright2024,CloudSecurityAlliance.Allrightsreserved.2AcknowledgmentsReviewersLeadAuthorsPhilAlgerIlangoAllikuzhiSiahBurkeBakrAbdouhMarcoCapotondiVinayBansalDanieleCattedduVijayBolinaKenHuangBrianBrinkleyAnupamChatterjeeContributorsJasonClintonAlanCurranMarinaBregkouSandyDunnVidyaBalasubramanianDavidGeeAvishayBarZackHamiltonMonicaChakrabortyVicHargraveAntonChuvakinJerryHuangRicardoFerreiraRajeshKambleAlessandroGrecoGianKapoorKrystalJacksonRicoKomendaGianKapoorVaniMittalKushalKumarJasonMortonAnkitaKumariAmeyaNaikYutaoMaGabrielNwajiakuDannyManimboMeghanaParwateVishwasManralPrabalPathakJesusL...