Analyze AI identity risks using Microsoft Defender XDR: Review

After hearing all the buzz about securing AI, I was hoping for a practical look at the topic, something beyond high-level talks. What I got was a 20-minute guided tour of a specific feature in the Defender XDR portal, and most of that time was just clicking 'Next.' The module is a series of screenshots showing how to discover 'AI Agents' and view their 'Attack Paths.' It’s all very specific to someone who already lives inside the Defender ecosystem.

There's no lab. You don't get a sandbox environment to try this out; you just look at pictures of the interface. The 'exercise' is just reading text descriptions. Then you get to the 'Knowledge Check' at the end, which was a letdown. One question was literally, 'Which portal do you use to discover AI agents?' The answer, of course, is the Microsoft Defender portal—the entire subject of the module. It feels less like a check of my understanding and more like a check to see if I was paying attention to the product names.

I get that these modules can be quick updates, but calling it training feels like a stretch. It’s a product announcement turned into a slideshow. It did introduce me to the term 'blast radius' for an AI agent, which was new, but I didn't get to calculate one or analyze a real scenario. I just saw a screenshot of where that information would appear if I had the right enterprise license and an active incident.