Default Deny: Hardening Your Cloud for Agentic AI

AI agents can rapidly test thousands of permission paths, creating security risks that traditional controls may miss. This webinar demonstrates how developers using agentic AI in real world scenarios can unintentionally create unauthorized roles and access points while attempting simple tasks. A live AWS environment is used to show how these situations occur and why existing RBAC models often fail to detect or prevent the resulting overprivilege.

The session explains how cloud native IAM guardrails can help manage these risks and why additional controls may still be required. It also presents the role of Cloud Permissions Firewall as a final defensive layer and shows how automated Default Deny policies can block agents from creating new identities or ingress points.