Politics & Economics

The AI Act passed: the EU is the first to regulate artificial intelligence

May 2024
By Alessandro Caruso

The AI Act is part of the European law. The Council of the European Union today definitively approved the EU law on artificial intelligence, which had been debated for months. The direction of the legislation is clear: the greater the risk of causing harm to society, the stricter the rules will be. This is in fact the world’s first systemic legislation on the subject, and so the hope of the European legislators is that it will become a global standard for the regulation of artificial intelligence, a field of technology in which the EU has lagged behind the US and China.

The new law aims to promote the development and adoption of secure and reliable AI systems in the EU single market by public and private actors. At the same time, it aims to ensure respect for the fundamental rights of EU citizens and to stimulate investment and innovation in the field of artificial intelligence in Europe. The law applies only to areas subject to EU law and provides exemptions, for example, for systems used exclusively for military and defence purposes as well as for research purposes.

The new law classifies different types of AI according to risk. Systems that present only a limited risk will be subject to very light transparency obligations, while those with a high risk will be authorised, but subject to a series of requirements and obligations to access the EU market. When the risk is unacceptable, on the other hand, bans are triggered: this is the case, for example, with manipulative techniques, predictive policing practices, and banned emotion recognition in the workplace and in schools. And again it is the case of facial recognition, the use of which is only permitted to law enforcement agencies and subject to strict conditions.

Another novelty is the chapter devoted to generative AI, which was inserted in the course of the work with the aim of providing an initial response to the spread of systems such as ChatGPT. General purpose AI (Gpai) models that do not present systemic risks will be subject to some limited requirements, e.g. regarding transparency, but those with systemic risks will have to comply with stricter rules, including model evaluation, systemic risk assessment and mitigation, and cybersecurity protection. The law promotes regulatory sandboxes and real-world-testing, set up by national authorities to develop and train innovative AI before it is placed on the market. Depending on the violation and the size of the company, fines will be imposed ranging from a minimum of 7.5 million euros or 1.5 per cent of turnover up to 35 million euros or 7 per cent of global turnover.

The AI Act applies only to areas covered by EU law and provides exemptions, e.g. for systems used exclusively for military and defence purposes and for research. After being signed by the Presidents of the European Parliament and the Council, the Act will be published in the Official Journal of the EU in the coming days and will enter into force twenty days after publication.