FROM OUR CORRESPONDENT
BRUSSELS – “Historic day”, everyone in Strasbourg repeats because they are the first rules in the world on the use of artificial intelligence and will set a global standard. The plenary of the European Parliament approved in Strasbourg The rules on AI with 523 votes in favour, 46 against and 49 abstentions. The Union reconfirms itself as a point of reference for the regulation of the digital economy, after the introduction of the General Data Protection Regulation (GDPR), the Digital Markets Act and the Digital Services Act.
“We are regulating as little as possible, but as much as necessary!”, commented the Commissioner for Internal Market and Digital Thierry Breton. “We have managed to put human beings and European values at the center of the development of artificial intelligence”, exulted Brando Benifei (Pd), co-rapporteur on the AI Act of the Internal Market Committee of the EU Parliament. While the co-rapporteur of the Civil Liberties Commission, the Romanian liberal Dragos Tudorache, underlined that the EU has “forever linked the concept of artificial intelligence to the fundamental values that form the basis of our societies”. The provisional agreement between Parliament, Council and EU Commission was reached on 9 December after 38 hours of negotiations, during which the negotiations on the use of AI for police purposes were the most intense.
AI developers and operators are given clear requirements and obligations regarding specific uses based on possible risks and level of impact. The regulation also sets limits, limiting the use of biometric identification systems by law enforcement agencies and introduces rules to combat the manipulation and exploitation of users’ fragilities. For example, artificial or manipulated images and audio or video content (so-called “deepfakes”) will have to be clearly labeled as such and there are specific obligations for the uses of AI in democratic processes such as in the case of systems used to influence elections. Furthermore, consumers will have the right to lodge complaints and receive relevant explanations.
The regulation still needs to be subjected to final verification by lawyer-linguists and should be definitively adopted within this legislature. But it has yet to be formally approved by the Council. It will enter into force twenty days after publication in the Official Journal of the EU and will begin to apply two years later, with some exceptions. Prohibitions on prohibited practices will apply starting six months after entry into force.
Commissioner Breton explained that «throughout the process» of negotiations the EU institutions have «resisted particular interests and lobbies that asked to exclude large AI models from the regulation. The result is a balanced, risk-based and future-proof regulation.” Now it becomes “fundamental – Benifei underlined – implementation together with voluntary compliance on the part of companies and institutions, which will have to apply the law gradually”: “There is a roadmap on compulsory law. We will start with bans, which is very important for us because we want to make it clear that some cases are prohibited and will not be allowed in Europe. The ban will come into force by the end of this year, the rest will follow.”
Applications that are dangerous for citizens such as biometric categorization systems that refer to sensitive personal data, such as religious beliefs, political or sexual orientation, and the extrapolation of facial images from the internet or CCTV systems are prohibited. closed to create facial recognition databases. Emotion recognition systems in the workplace and in schools and predictive policing practices will also be banned. But law enforcement will be able to use remote biometric recognition in real time to search for a missing person or to prevent a terrorist attack.