AI Act: Which obligations are already in force?

EU Regulation 2024/1689 (AI Act) entered into force on August 1, 2024, but will become applicable in stages.

This choice was driven by the desire to introduce the legislation gradually, guiding market operators through this change and avoiding negative impacts caused by haste.

The obligations will become applicable according to the following timeline:

• February 2, 2025 → training obligations for all entities using AI and a ban on the use of unacceptable-risk AI.
• August 2, 2025 → supervisory authorities are appointed, all obligations for GPAI models apply, and penalties for AI systems are enforced.
• August 2, 2026 → all obligations for high-risk AI systems and penalties for GPAI models apply.
• August 2, 2027 → obligations relating to AI systems that are safety components of other systems apply.

Regardless of the direct use of AI, training should be the first best practice, as well as a mandatory requirement for those already using AI systems. In fact, we often experience the consequences of artificial intelligence without even realizing it. Training makes us more aware, protects us, and guides us in our professional and personal choices.

For those already using AI, in addition to training, the first step to manage it effectively and compliantly is to map their own AI systems. Subsequently, it will be possible to identify which risk categories they belong to and act accordingly.