Secure distributed multi-agent systems with Model Armor | Google Cloud Live
Model armor Multi Agent system security Google cloud security Prompt injection defense Sensitive data protection Pii redaction Terraform policy as code Ai agent security Red team testing Defense in depth
This session teaches developers how to secure multi-agent AI systems using Google Cloud's native security tools, focusing on implementing Sensitive Data Protection to redact PII before it reaches agents, configuring Model Armor to intercept malicious prompt injections, and running Red Team scenarios to stress-test defenses. It covers Policy-as-Code implementation with Terraform for consistent security posture across environments, designed for developers building production AI systems who need to implement defense-in-depth strategies.
