Skip to content
Learn · Guide · For teams

The EU AI Act: A Field Guide

What the EU AI Act actually requires, in plain language. Risk tiers, who's on the hook, and the deadlines that matter.

The short version

Most EU AI Act explainers are either a press release or a 100-page PDF. This is the version a busy team can act on. The Act sorts AI systems into four buckets by risk: **prohibited**, **high-risk**, **limited-risk** (transparency duties), and **minimal-risk**. Almost everything you build is minimal- or limited-risk — but the obligations jump sharply the moment a system touches hiring, credit, education, biometrics, or safety. Three things teams get wrong: assuming "we just use an API" means it doesn't apply; missing the transparency duty to tell users they're talking to AI; and treating compliance as a one-time gate rather than ongoing monitoring. The full guide gives you a one-page decision tree to find your tier, the obligation checklist per tier, the phased deadlines, and a worked example for a mid-market company.
The EU AI Act: A Field Guide · SDEN