《在金融服务领域构建注重隐私的生成式人工智能应用.pdf》由会员分享,可在线阅读,更多相关《在金融服务领域构建注重隐私的生成式人工智能应用.pdf(21页珍藏版)》请在三个皮匠报告上搜索。
1、 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.I N D 3 4 5Building privacy-aware generative AI applications in financial servicesPiyali Kamra(she/her)Sr.Solutions ArchitectAWSJigna Gandhi(she/her)Sr.Solutions Arc
2、hitectAWS 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.AgendaNeed for privacyaware Generative AIPersonas and scenariosAWS services to built privacy layerReference ArchitecturesDemoSummary&resources 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.Need for Priv
3、acy-Aware Generative AIProtect sensitive dataPreserve trustPersona based insightsRegulatory requirementsPersonalized scale 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.What challenges do you face in sharing sensitive documents today?2025,Amazon Web Services,Inc.or its affiliate
4、s.All rights reserved.AWS Services to Build Privacy LayerAWS Identity and Access Management(IAM)Amazon CognitoAmazon Verified PermissionsAmazon BedrockGuardrailsAWS OrganizationsService Control PoliciesAmazon BedrockAutomated Reasoning 2025,Amazon Web Services,Inc.or its affiliates.All rights reserv
5、ed.Amazon Verified PermissionsCan I access this?ApplicationAmazon Verified PermissionsCheck policiesPermissions-as-PoliciesCedar Policy Examplepermit(principal,action,resource)when principal.role=SeniorAnalyst&principal.sox_certification=true 2025,Amazon Web Services,Inc.or its affiliates.All rights
6、 reserved.How you compare different access level in personas-region,line of business,clearance level?2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.Amazon Bedrock GuardrailsUser inputFinal responseFM responsesFM inferenceAmazon Bedrock invocationsGuardrailsWord filtersDenied topi