Risk levels are legal labels, not a mood

A meeting can feel risky and still sit outside a legal category

Teams say "high risk" when a project is new, expensive, or embarrassing if it fails. That is a business judgment. The EU AI Act uses risk categories with defined meanings, including a short list of prohibited practices and a list of uses that carry extra duties, such as certain uses in hiring, education, credit, and essential services. A feeling in a workshop does not put you on that list, and staying calm in the workshop does not take you off it.

The Act is European Union law. It can reach some organizations outside the EU, for example when they place a system on the EU market or when the output is used in the EU. It does not apply to every business in the world. Our article "The EU AI Act in plain English" walks through that scope and the dates. Dates in older summaries are easy to get wrong, so check the current text before you rely on a year.

Other countries do not borrow the same ladder

A use that sits in one EU category may be treated under a different statute elsewhere, or under general consumer, privacy, or sector rules. Sector rules for health, finance, or employment can matter even when an AI-specific law does not. Stack the questions. Do not assume one ladder answers all of them.

Use the word you mean

In internal notes, say "important to us" when you mean business impact, and reserve "high-risk" for the moment you are discussing a defined legal category. Mixing the two makes later reviews look more certain than they were.

This is general education, not legal advice. Classification of a real system needs the actual use, the actual market, and qualified advice.