Last Tuesday, 24th October the 4th political trilogue on the AI Act took place. Time for another update. Check the graphic below for an overview of the upcoming & previous sessions as well as all approved batches.
It was a very intense meeting, lasting almost until midnight. Yet, the European Parliament and the EU2023ES have made further progress: in particular by finally agreeing on the mechanism of how to classify high-risk AI-systems in Article 6.
At the same time, there are still many lines open or parked. The big stumbling blocks remain unsolved: (1) the prohibitions in Article 5, (2) foundation models & generative AI, and (3) the AI Office. More work is also needed for topics such as FRIA, the definition of AI, Annex III or real-world testing.
Too much for the five weeks until the 6th December, the date of the final political trilogue? It is tight but not impossible. Axel Voss and I remain fully committed. We will give everything to close the political negotiations under Spain this year as we do firmly believe that delaying the AI Act's adoptions after the 2024 EU election is NOT a wise move. The EU dominates the global discourse on how to regulate AI and could shape the upcoming international standards, creating a competitive edge for our companies and researchers.
As Better Regulation proponents, we will however not allow that speed is trumping quality. In the end, only a coherent AI Act will allow the development and deployment of trustworthy AI made in the EU. This is the outcome, we will continue to fight for!