Ambassadors from the 27 EU countries have unanimously approved the world's first sweeping rulebook for AI. The AI Act is the first of its kind to be adopted and aims to make AI safe and respect the fundamental rights of EU citizens.

In December, EU officials reached a consensus on the key sticking issues of the AI Act, a landmark initiative that regulates AI based on its potential to do harm. Due to the law's complexity, it took over a month to tweak and complete.

Why does it matter?

At a technical meeting on 24 January, the Belgian president of the Council of EU Ministers presented the final version of the legislation. Many member nations immediately expressed reservations about the text since they did not have enough time to review it. Most objections were addressed by 2 February, when the Committee of Permanent Representatives met to adopt the AI Act. However, because several European heavyweights, including France, still opposed certain aspects of the text, approval from EU ambassadors was not clear until the very last minute.

What's next?

The bloc's institutions will now move into the implementation phase, where the focus will be on enacting and enforcing the new regulations. Individual EU nations can still influence how the AI Act is executed since the European Commission has to adopt about 20 'acts of secondary legislation.'

Paris, for instance, wants the AI Act to be implemented in a way that balances transparency and trade secret protection, avoids overburdening companies with high-risk obligations, and ultimately does not hinder the development of a competitive AI industry. Additionally, the future 'European AI Office' designed to monitor AI models will have to be staffed by national specialists.