EU countries agree to changes to artificial intelligence regulation

European Union countries, together with MEPs, have agreed to changes to the regulation of artificial intelligence aimed at simplifying the current legislation. The aim of the decision is to reduce the bureaucratic burden on business and increase the competitiveness of European companies in the fight against technology giants from the US and Asia.

According to Reuters, the updated approach involves postponing some of the strict requirements, and some provisions have even been completely abolished. Despite this, the EU's legislation on artificial intelligence remains one of the strictest in the world.

The participants in the agreement emphasize that the changes give developers more time to adapt to the new rules. In addition, administrative costs for companies working in the field of artificial intelligence are expected to decrease.

One of the key decisions was the postponement of inspections of high-risk AI systems. In particular, control over technologies such as biometric systems or tools for law enforcement activities has been postponed until December 2027.

Separate provisions concern the protection of citizens’ personal rights. From December 2026, the EU will ban the use of artificial intelligence to create intimate deepfakes — sexual images or videos without a person’s consent.

Also, from December 2026, a requirement to label content created by artificial intelligence will come into force. All such materials must have special marks or watermarks, which will allow users to clearly distinguish them from real content.

At the same time, legislators decided to reduce regulatory pressure on industry. Industrial equipment was excluded from the new rules, since it already has separate industry standards governing the safety and use of technologies.