Enterprise AI Security Workshop

Overview

This workshop treats AI security as a systems problem, not a checklist. It's built around one running example — a FinOps multi-agent application — that each chapter layers its controls onto, so identity, data protection, platform hardening, and runtime observability are shown solving the same architecture's problems rather than four disconnected topics. Two deep-dive chapters (Model Context Protocol security and RAG security) go further into two specific technical areas participants consistently ask about.

Who this is for

• CISOs evaluating how existing security programs need to change for agentic AI.
• Security architects designing identity, data, and runtime controls for AI systems.
• Enterprise architects deciding where AI security controls should live in an existing platform.
• AI engineers who need to understand the security implications of the systems they're building.

Agenda

Time	Session	Chapter
9:15 – 10:00	Evolving Trust Boundary	Ch. 1–2
10:00 – 11:00	Setting up Agent Governance — Agent ID	Ch. 3
11:00 – 12:00	Shadow AI & Data Security	Ch. 4
12:00 – 1:00	Lunch break	—
1:00 – 2:00	Test your AI hacking skill (capture the flag)	—
2:00 – 2:30	Setting up secure foundation	Ch. 5
2:30 – 2:45	Break	—
2:45 – 3:30	Operating securely, runtime protection, and Q&A	Ch. 6

MCP Security and RAG Security (Chapters 7–8) are technical deep dives delivered as follow-up material.

Chapters

1. 1. The Shift to Agentic AI · ~8 min — trust boundary evolution and the 8-domain Agentic AI Security Framework.
2. 2. The FinOps Reference Architecture · ~7 min — the running case study and agent anatomy.
3. 3. Managing AI Identities · ~9 min — Entra Agent ID: Register, Govern, Protect.
4. 4. Data Security and Shadow AI · ~9 min — the Samsung case study and Purview's agent protections.
5. 5. Securing the AI Platform · ~10 min — the Amazon Q incident, content safety, and automated red teaming.
6. 6. Operating AI Securely at Scale · ~11 min — runtime detection, telemetry design, and Agent 365.
7. 7. Deep Dive: MCP Security · ~9 min — the OWASP MCP Top 10 and a reference Azure-native architecture.
8. 8. Deep Dive: RAG Security · ~10 min — a 7-stage RAG pipeline security walkthrough.

→ Glossary of terms used throughout this workshop