Leading MEPs push back on EU AI rulebook’s law enforcement section

Leading Members of the European Parliament (MEPs) have made a counter-proposal on the law enforcement chapter of the AI rulebook.

 People, Person, Indoors, Auditorium, Hall, Theater, Architecture, Building, Classroom, Room, School, Crowd, Audience, Lecture, Lecture Hall

Dragoș Tudorache and Brando Benifei, European Parliament co-rapporteurs for the AI Act, circulated a compromise document addressing the controversial issue of law enforcement’s use of AI. Several EU member countries are advocating for more flexibility for their security and police forces. Most MEPs are opposed to this to protect fundamental rights.

The counter-proposal provides exclusions for conducting a focused search for specific victims of a predetermined list of major crimes and locating a suspect in connection with these crimes.
MEPs decided to put AI-powered polygraphs and other comparable technologies on the high-risk list, but only when they are used directly by law enforcement and in compliance with EU and national legislation.
The proposal also removes law and migration enforcement uses of AI systems like crime analytics, deepfake detection, and migration trend predictions. Except for travel document verification, any AI application for border control would be deemed high-risk.

Why does it matter?

The AI Act seeks to regulate AI based on its potential to harm humans and their fundamental rights. The aim is to ensure that AI systems are safe, transparent, traceable, non-discriminatory, and environmentally friendly. This counter-proposal by prominent MEPs comes as the world’s first comprehensive AI law is entering its final phase of intense negotiations between the EU Parliament, Council, and Commission in so-called trilogues. The impact of the legislation could be felt well beyond the bloc’s borders, in what is often described as the ‘Brussels effect’. EU institutions still need to agree on how to handle the most powerful and advanced ‘foundation’ models and the specific provisions for the sensitive law enforcement domains.

Led by France, several member countries are lobbying for a broad national security exemption, including AI technologies designed or used solely for military objectives. However, as recommended by the co-rapporteurs, the consensus among all political parties in the European Parliament is for a more restrictive approach. The final version of the EU AI rulebook will likely be the outcome of further negotiations and compromises in the weeks and months to come.