
As AI adoption accelerates, most organizations are securing it the wrong way. Real-world AI security incidents rarely start with rogue models or agents — they start with stolen credentials, over-privileged access, and identity misuse.
In this webinar, Ian All, CTO at Permiso, breaks down why AI security today is fundamentally an identity security problem. Drawing from real breaches like the SalesLoft incident and frontline research from Permiso’s P0 Labs, Ian shows what’s actually happening in the wild — and where defenders should focus now.
Attendees will learn how to:
Understand AI security through the lens of identity, not just posture and configuration
Distinguish between AI users, AI builders, and AI agents — and why each carries different risk
Detect credential abuse tied to AI services, OAuth tokens, and non-human identities
Gain runtime visibility into how AI is really being used across cloud, SaaS, and endpoints
As AI usage expands faster than governance, identity remains the common thread attackers exploit. This session shows how an identity-centric approach brings clarity, context, and control to modern AI risk.