Who Decides Your AI Is 'High-Risk'?

Here's something people get wrong: "high-risk" under the EU AI Act is not a vibe. It's a list.

The Act names the areas where AI counts as high-risk: hiring and firing decisions, school admissions and exam grading, credit scoring, access to public benefits, law enforcement, and a few more. If your AI system operates in one of these areas, or acts as a safety component in regulated products, it's high-risk. Full stop.

That means your first compliance task isn't philosophical. It's clerical. Take the list, walk through what your AI actually does, and check for a match.

My rule of thumb: if your AI makes decisions about someone's job, education, money, or freedom, assume it's high-risk until you've proven otherwise. It's easier to do the homework than to explain later why you skipped it.

I explain frameworks; I don't give legal counsel.