New AI requirements come into force in the EU

From August 2, 2026, the main body of the AI Act - the Artificial Intelligence Act - comes into effect in the European Union. The new rules set requirements for developers, providers, and companies using AI systems.

The law provides for a risk-based approach: the greater the potential threat a system poses to people's rights and safety, the stricter the requirements for its use will be.

Users must be warned about the use of AI

One of the key novelties will be transparency requirements. People must be informed if they are interacting with a chatbot or other AI system, unless it is obvious from the context.

Developers must also make it possible to detect images, audio, video, and texts generated or altered by AI. Such materials must carry technically detectable labeling in a machine-readable format.

In addition, users must be warned about the display of deepfakes, the use of emotion recognition systems, or biometric categorization. AI-generated texts on matters of public interest must also be labeled, if they are published without human verification or editorial oversight.

Some rules for high-risk systems delayed

However, not all provisions of the AI Act come into force at the same time. According to the updated calendar of the European Commission, requirements for high-risk systems used in employment, education, lending, insurance, law enforcement and other sensitive areas are to begin applying from December 2, 2027.

For AI systems embedded in medical devices, cars, toys, elevators and other products already governed by separate EU rules, the transitional period will last until August 2, 2028.

The prohibition of the most dangerous practices started applying back on February 2, 2025. Specifically, systems of social scoring, mass untargeted scraping of facial images from the internet, manipulative technologies, and certain cases of biometric identification of people in public places are banned in the EU.

Starting from August 2025, rules for general-purpose AI models, including large language models capable of performing a wide range of tasks, are also in effect.

Penalties provided

Companies using prohibited AI systems may face fines of up to €35 million or up to 7% of their worldwide annual turnover – whichever is greater.

Non-compliance with other AI Act requirements is punishable by fines of up to €15 million or up to 3% of global annual turnover. For small and medium-sized enterprises, the smaller of the two amounts will apply.

The AI Act entered into force on August 1, 2024, but its provisions are being introduced in stages. The regulation applies directly in all member states of the European Union.

Based on materials from: European Commission, EUR-Lex