Today's blog post delves into the latest development from the EU concerning the regulation of Artificial Intelligence—the AI Act (AIA). Over the next three weeks, our weekly posts will centre around the main requirements of the AIA and explore potential gaps in this comprehensive legislation. Minor revisions to the text will be made in the coming months, necessitating a wait before reviewing the final wording of the regulation.

The AI Act applies to organisations introducing AI systems to the EU market, deploying them in service, or utilising them within the EU. The defined AI systems are described as “a machine-based system designed to operate with varying levels of autonomy and that may exhibit adaptiveness after deployment and that, for explicit or implicit objectives, infers, from the input it receives, how to generate outputs such as predictions, content, recommendations, or decisions that can influence physical or virtual environments.”

Two notable elements of the definition merit attention. Firstly, it encompasses systems where AI operates autonomously and under human guidance, contrasting with Article 22 of GDPR, which covers only decisions made solely through automated means. Secondly, it includes systems that adjust based on provided information and those that learn to generate outputs based on received data.

Thus, the AI systems' definition is notably broad, ensuring flexibility and adaptability. Exemptions from the regulation include systems already subject to harmonisation legislation, as well as those used exclusively for military, defence, or national security purposes, scientific research, development, and purely personal non-professional usage.

A highly discussed exception to the AI Act's applicability pertains to AI systems utilising free and open-source software, which will be addressed in the final blog post, while the next one will focus on the Regulation's classification system.

Before concluding, the timeline of the AI Act shall be discussed. Note that these dates are approximate. In April 2024, the final text of the AI Act will be published. A transition period of six months commences, and in October 2024, the prohibition of certain AI systems comes into force. In April 2025, provisions regarding general-purpose AI (GPAI) systems will come into effect, and finally, in April 2026, the remaining provisions will be enforced.

Unveiling the AI Act - written by Maria Mot