This session explores why identity is the new security perimeter in a world of copilots, chatbots, and intelligent apps.

We will walk through common identity and access threats in AI-powered scenarios, including data exfiltration, prompt and session abuse, and account compromise, and show how strong authentication, conditional access, and least-privilege permissions can mitigate them.

Using real world scenarios and reference architectures, we will demonstrate how to design identity-first, Zero Trust-aligned patterns for AI applications and services.

The intended audience is cloud and application developers, solution architects, security engineers, and decision makers who are building or securing AI-driven workloads and need practical guidance on identity controls.

By the end of this session, attendees will be able to: identify key identity risks introduced by AI workloads, apply concrete patterns for protecting data and user accounts with modern identity platforms, and evaluate their own AI architectures against a set of actionable best practices.