Most mid-market companies believe a signed AI policy already covers them. Inspections under the EU AI Act started on 2 August 2026, carried out by AESIA in Spain and equivalent national authorities across the rest of the EU. What they actually check is different: training records tied to named people, not a policy document sitting in a drive somewhere.
A free guide walks through exactly what closes that gap, including what the fines actually look like if you don’t. It applies to more companies than the headline suggests.
Not only the teams building AI
If your company runs on commercial tools such as Copilot, ChatGPT, or Gemini, and you’ve assumed Article 4 mostly applies to businesses building their own AI systems, this is the part worth reading closely. The obligation reaches your team too, on top of whatever your AI providers already handle on their side. It’s an easy gap to miss, even for companies that are already deep in compliance work elsewhere.
What the guide actually covers
It walks through what an inspection really checks and the three things that look like compliance and are not: a policy-only company, a one-off course, or access without training. It closes with what a defensible training programme actually looks like, in what order and at what cost.
Download the free guide to becoming Article 4 compliant under the EU AI Act.
If you’d rather start with a fast number instead of a full read, our two-minute diagnostic tells you where your company’s risk actually sits first.