Blockade in Brussels: Negotiations on EU AI rules failed for now

The trilogue negotiations on the future of the AI Act in Brussels ended without results early Wednesday morning. The teams from the EU Parliament, the member states, and the Commission parted ways after hours of talks without reaching an agreement on the so-called Digital Omnibus. This pushes the goal of reforming parts of the AI regulation and banning AI-powered “nudifier” apps further into the distance.

The continuation of the talks has been postponed indefinitely for now, reports Politico, citing diplomats and EU officials. The new rules for high-risk AI applications were actually supposed to come into force in August. The failure of the negotiations leaves the industry with legal uncertainty.

A sticking point in the negotiations were demands from the center-right EPP group in parliament, which Chancellor Friedrich Merz (CDU) shares. They are pushing to largely exclude industrial AI applications – for example in medical technology or mechanical engineering – from the scope of the AI Act.

This is intended to protect German heavyweights like Siemens or Bosch from double regulation and secure their competitiveness. Proponents of this “sectoral exit” argue that these products are already subject to strict sectoral laws and that additional requirements from the AI Act create unnecessary bureaucracy.

Political exchange of blows

However, this request is meeting with resistance from the Greens and Social Democrats, as well as several EU countries. Dutch MEP Kim van Sparrentak spoke of a “coup” by the German wing of the EPP. She warned that the industry would be plunged into legal chaos by this blockade. Green rapporteur Sergey Lagodinsky accused Merz of holding the European legislative process hostage.

Critics consider it particularly serious that the blockade has also stalled socio-political initiatives such as the ban on sexualized deepfakes. Lagodinsky accuses the CDU and CSU of preventing the planned ban on undressing apps.

The TÜV Association warns against abandoning the horizontal approach of the AI Act. The uniform regulation of high-risk AI across different product groups is the core of the regulation. A retreat into sectoral individual regulations would lead to a long regulatory vacuum and a patchwork. Instead of reducing bureaucracy, different standards and procedures would emerge, weakening Europe as an industrial location. Furthermore, such a step would jeopardize the opportunity to establish European standards as a global benchmark for trustworthy AI.

Warning of insufficient fundamental rights protection

AlgorithmWatch criticizes that the AI Omnibus is being rushed through anyway, without adequately considering the consequences for fundamental rights. The organization demands that a ban on sexualizing deepfakes should not fail due to technical definitions of “intimate body parts.” The explicit consent of those affected must be decisive. The delay in negotiations offers room for improvements. At the same time, the risk is growing that no protective mechanism will be adopted at all.

Ben Brake, Director of the platform lobby DOT Europe, called the breakdown of talks a setback for the digital economy. He urged a swift return to the negotiating table. The Commission also expressed concern: Vice-President Henna Virkkunen is pushing for a quick agreement, but one that should not stand in the way of high-quality legislation.

(vbr)