The EU Regulation on Artificial Intelligence (1689/2024) published in the OJEU

Published in the Official Journal of the European Union, Regulation 1689/2024 on artificial intelligence (so-called AI Act), which introduces a system of harmonized rules for the development and marketing of artificial intelligence (AI) systems within the European Union.

The aim of the Regulation is to reconcile innovation and market competitiveness by ensuring "a high level of protection of public interests, such as health and safety and the protection of fundamental rights, including democracy, the rule of law and environmental protection".

The measure adopts a risk-based approach, distinguishing AI systems into different categories based on the type of risk to user safety and to the respect of citizens' fundamental rights, with specific obligations on suppliers and "professional" users (so-called deployers) for each category.

Specifically, prohibitions are provided for the so-called “unacceptable” risks of AI systems, with a ban on their introduction into the common market (art. 5), or the marketing of so-called high-risk ones (arts. 6 and 7) can only take place after compliance with a series of requirements, including the adoption of a risk management system (art. 9), the drafting and updating of technical documentation (art. 10), the preparation of technical measures that guarantee the possibility of understanding the output of the device (art. 13) and of effectively supervising it (art. 14).

Instead, for limited-risk systems, compliance with certain transparency obligations will be sufficient.

In the event of failure to comply with the provisions established by the Regulation, art. 99 obliges Member States to introduce “effective, proportionate and dissuasive” sanctions, which may also include warnings and non-pecuniary measures.