本课程回顾了 Model Armor 的基本安全功能,并让您能够使用该服务。您将了解与 LLM 相关的安全风险,以及 Model Armor 如何保护您的 AI 应用。
As enterprise teams transition to autonomous AI agents, security evolves toward proactive, identity-centric protection. In this course, you will implement a zero-trust security architecture for Cymbal Banking's underwriting platform using the Gemini Enterprise Agent Platform. By deploying a regional Egress Agent Gateway, configuring Model Armor and Semantic Governance (SGP) safety templates, assigning cryptographic SPIFFE identities to agent runtimes, and declaring a VPC-SC perimeter in Terraform, you will protect database boundaries and manage tools with production-grade governance.
了解如何在 Google Cloud 上保护和治理 AI 智能体。本课程提供了一个实用的路线图,可帮助您保护智能体记忆,通过 IAM 和 VPC Service Controls 实施严格的访问权限控制,并部署实时对话防护措施,从而确保企业部署安全合规。