The AI Act has started
Feb 12th, 2025 by Jesper Kråkhede
On the 2nd of February the first parts of the EU AI Act started to apply. Many in management have heard of it, few have read it. Here is what matters now.
The first rules are prohibitions. Certain uses of AI are now banned in the EU, for example systems that manipulate people in harmful ways, social scoring and some forms of emotion recognition in the workplace. Most organisations don´t do any of that, but you need to be sure.
The other requirement that applies now is easy to miss: AI literacy. Organisations that provide or use AI systems must make sure that their staff have sufficient knowledge of AI to use it responsibly. That covers a lot more organisations than the prohibitions do.
More is coming. Rules for general purpose AI models apply from August this year, and most of the requirements for high risk systems follow in the coming years.
So what should the board do? First, find out where AI is used in the organisation, including the AI built into products you already buy. Second, check that nothing falls under the prohibitions. Third, start a basic AI training programme for staff. Fourth, decide who owns AI governance. It shouldn´t be the IT department alone. It touches legal, HR, risk and the business.
The AI Act isn´t only a compliance exercise. It is a reason to understand what you are actually doing with AI.
You can´t govern what you haven´t found.