Agent Defense: Securing AI Agents - 17 de junio de 2026 - TecnoWebinars.com

The frontier of enterprise AI involves autonomous agents capable of executing complex tasks, interacting with internal databases, and communicating with other agents. This paradigm shift introduces unprecedented security and authorization challenges. This advanced session focuses on the architecture and protocols required to secure AI agents, ensuring they operate within strict boundaries of trust, observability, and least privilege. Key Topics Covered: • Agent Identity & Access Management (IAM): Provisioning, tracking, and managing robust non-human identities for autonomous agents operating within Google Cloud environments. • Agent-to-Agent (A2A) Security: Securing communications and establishing trust boundaries and cryptographic verification between interacting AI agents. • Model Context Protocol (MCP) Servers: Best practices for securely deploying MCP servers to expose enterprise data and tools to AI agents without compromising access controls. • API Gateways & Guardrails: Enforcing strict authorization, rate limiting, and behavioral guardrails for agent actions to prevent privilege escalation and unauthorized workflow execution.