With the European Commission’s long-awaited Guidelines on prohibited AI practices now published, enforcement under the EU AI Act is moving closer to reality.

Article 5 leaves no room for mitigation or balancing tests. If a system falls within its scope, it must not be deployed.

The real board-level risk? Many prohibited practices do not appear as intentional design choices. They emerge gradually – through optimisation goals, behavioural analytics, vendor tools or revenue pressure.

These risks are not theoretical. They can surface across e-commerce, financial services, insurance, HR and marketing systems.
For boards, the key question is no longer only what AI is intended to do, but whether – and how – it influences human decision-making in practice.

Article 5 exposure is now a board issue – not just a compliance checklist item.

In our latest briefing, Andrea Belényi and Endre Várady outline five prohibited AI practices that boards should review now, together with practical governance actions companies may wish to consider before enforcement accelerates.

Read the briefing here.

Stay connected for more expert insights – follow us on LinkedIn.