Transforming Identity Protection: Innovating with AI and Attack Paths

In this insightful presentation from the DEF CON 33 Cloud Village, speaker Filipas addresses the escalating challenges of securing machine identities across modern multi-cloud architectures. Opening at 00:00, Filipas outlines how rapid cloud adoption has resulted in non-human identities vastly outnumbering human users, creating complex and fragmented security boundaries.

At 03:00, he examines the fundamental anatomy of machine identities—including IAM service accounts, API tokens, IAM roles, and automated workload credentials across AWS, GCP, and Azure. At 05:01, he analyzes common misconfigurations in cloud trust policies, showing how over-permissive role assumptions and privilege sprawling enable attackers to execute lateral movement.

He demonstrates attack path modeling at 12:00, illustrating how graph analysis can map hidden identity relationships and privilege escalation vectors. He concludes at 18:00 by advocating for continuous entitlement auditing, least-privilege automation, and AI-driven path disruption to isolate and protect critical cloud infrastructure. (AI summary — Gemini Flash)