In the era of autonomous agents, a standard safety filter is no longer enough. When we give agents the power to reason, access data, and execute tools, we open a massive new attack surface. How do we prevent prompt injections, PII leakage, and jailbreaks in a multi-agent system?

In this session, we will explore how to secure agentic workflows using Google Cloud Model Armor. We will explore Intent-based Protection, Data Sanitization & Redaction, Governance at Scale.

It’s time to move from experimental AI to production-ready systems. Let’s see how to shield your models and protect your users without killing the 'magic' of the agent.