The European AI Regulation

In June 2024, European lawmakers signed the regulation on artificial intelligence (AI Act).

The AI regulation is the first binding, horizontal legislation for AI worldwide and establishes a common framework for the use and offering of AI systems in the EU. The new law introduces a classification for AI systems with different requirements and obligations, tailored to a "risk-based approach."

AI systems that pose an 'unacceptable risk' are strictly prohibited. While a wide range of 'high-risk' AI systems—those potentially impacting health, safety, or fundamental rights—are permitted, they are subject to strict requirements and obligations to access the EU market. Thor is designed to meet the highest regulatory standards: systems with limited risk comply with all transparency and information requirements, while those posing only minimal risk are managed with full compliance in mind.

The regulation also introduces specific rules for General-Purpose AI (GPAI) models. More stringent requirements apply to GPAI models with 'systemic risk capabilities' that could impact the internal market or pose systemic risks.

The AI regulation was published on July 12, 2024, in the EU Official Journal and came into effect in August 2024.