Every AI Agent Needs an Identity, and a Human Behind It
Extending identity security to autonomous AI
Peter Giesa – Security Lead, Cloudworks
As organizations rush to adopt AI, the security conversation has sprawled across data leakage, model integrity, prompt injection, shadow AI, and compliance. For identity teams, one shift matters most: AI agents no longer just respond, they plan, call APIs, use tools, and take autonomous actions on a human's behalf, at machine speed.
This session reframes AI security as an identity problem. We will walk the landscape of what needs securing, then focus where identity teams have real leverage: treating every agent as a first-class non-human identity with a named human owner, least-privilege scopes, a managed lifecycle, and continuous monitoring.
You will leave with a clear principle to act on: every agent needs its own identity, linked to the human it acts for, and what that means for your AI security strategy.
Bio:
Security Lead with 20+ years of experience and a background in IAM, endpoint management, and Zero Trust security, experienced in both technical architecture and leadership roles, and known for translating complex security challenges into practical, business-aligned solutions.