The four EU AI Act risk tiers

The EU AI Act sorts every AI system into one of four risk levels. Your level decides what the law asks of you — and how much work compliance takes.

• Prohibited

  This use is banned in the EU.

  What to do: You must stop it or redesign the system — no amount of paperwork can make it compliant.

• High

  Extensive duties: documentation, risk, oversight.

  What to do: The heaviest duties: a quality and risk-management system, technical documentation, human oversight, and conformity assessment.

• Limited

  Mainly transparency obligations apply to you.

  What to do: Mainly transparency — clearly tell people when they’re dealing with AI or AI-generated content.

• Minimal

  Very limited risk — light obligations.

  What to do: Light-touch. Mostly voluntary good practice; very few hard obligations.