The institutions of the European Union, which since Wednesday afternoon have been trying to finalize an agreement on the artificial intelligence lawhave already agreed on the rules to regulate the foundational models on which systems such as ChatGPTaccording to sources familiar with the negotiations cited by Efe.

The way to regulate this type of systems, which have exploded in the last year, has been one of the main points of friction between community institutions when agreeing on the EU artificial intelligence law, one of the first global regulations on this technology. Nevertheless, There are still “many things” to closeaccording to the sources consulted by the aforementioned agency.

The Council of the EU, the European Parliament and the European Commission began the meeting this Wednesday at 3:00 p.m. to try to agree on the regulations and gave themselves until this Thursday at 8:00 a.m. to close the agreement. However, this Thursday morning have decided to continue negotiations for longer.

The other major point of friction in the negotiations are the exceptions in the regulations to allow the real-time biometric surveillance in public spaces. European governments want it to be used to prevent crimes such as terrorism, sexual abuse or the protection of critical infrastructure, always with prior judicial authorization.

The European Parliament, on the other hand, opposes its use because it considers that biometric surveillance violates fundamental rights, although the rapporteur of the norm in the European Parliament, the social democrat Brando Benifei, has been open to allowing it if there are strong guarantees to protect the rights.