This course contains the use of artificial intelligence.

Autonomous AI agents don't just answer questions — they take multi-step actions with real tools, real data, and real consequences. That shift from chatbot risk to agent risk is where most governance programs fall short. This course closes that gap with a hands-on, practitioner-focused operating model for securing and governing AI agents in real organizations.

You'll work through one realistic company, Meridian Commerce, and its three agents, building a complete governance pack step by step: a capability inventory, an agentic-AI risk register, a tool-permission matrix, a control matrix, an evaluation checklist, an incident-response playbook, and an executive one-pager. Every risk is mapped to recognized framework language from the NIST AI Risk Management Framework, ISO/IEC 42001, and the OWASP Top 10 for LLM Applications.

Across 10 sections you'll learn to threat-model agents (prompt injection, tool misuse, data exfiltration, reward hacking, sandbox escape), design least-privilege and human-approval controls, test them with capability and safety evaluations, monitor agents at runtime with kill-switches and incident playbooks, run vendor due diligence, and stand up a lightweight governance operating model. No coding is required — optional technical labs use diagrams and control templates.

By the end, you'll be able to design, assess, approve, monitor, and govern AI agent deployments with confidence, and you'll walk away with a board-ready governance pack you can apply immediately.