Securing AI Agents

AI agents represent the next evolution of APIs, bringing new security challenges and attack vectors. This session examines real-world adversarial threats including prompt injection, tool abuse, and agent-to-agent escalation. Learn defensive strategies using rate limiting, encrypted endpoints, and Amazon Bedrock Guardrails. The session demonstrates how to build secure-by-default agents using Infrastructure as Code, enforcing deterministic behavior and Zero Trust principles. Essential knowledge for teams deploying autonomous agents in production environments.