IT Glossary
What is the EU AI Act?
The AI Act is the European Union regulation setting rules for artificial intelligence systems according to the risk they present.
The world's first comprehensive law dedicated to artificial intelligence came out of the European Union. The AI Act entered into force in 2024, with its obligations phasing in through 2026 and 2027. Its logic is a pyramid of risk. Unacceptable practices — manipulation, social scoring — are simply prohibited. High-risk systems, meaning AI used in recruitment, credit, education or medical devices, carry strict requirements for documentation, human oversight and data quality. Limited-risk applications mostly carry transparency duties, such as telling a user they are talking to a chatbot rather than a person. The point most companies miss is that the regulation does not only bind the people who build AI; it also binds the companies that merely use it. If you screen CVs with an AI tool, you already have obligations. With penalties reaching 35 million euro or seven per cent of global turnover, taking an inventory of the AI systems already in use across your business has stopped being optional housekeeping.
Let’s talk about your project
Message us on WhatsApp or send an email — you talk directly to a developer.
office@northdan.com · +40 752 070 247
Why it matters for your business
Penalties avoided through correct classification
Once you know which risk category each AI tool falls into, you meet exactly the obligations that apply — no gold plating, and no unpleasant surprises during an inspection.
An advantage in tenders and partnerships
Corporate clients and public institutions have started requiring AI Act compliance from suppliers. Having it early makes you the easier partner to sign.
AI projects built right from the start
Transparency, documentation and human oversight designed in at the outset cost a fraction of what it takes to retrofit them into a running system.
Frequently asked questions
We only use ChatGPT and a website chatbot — does the AI Act apply to us?
Yes, but lightly. For that kind of use the main duties are transparency, so visitors know they are talking to an AI, basic AI literacy training for the staff using the tools, and labelling generated content in certain contexts. The heavy obligations attach only to high-risk systems.
Which AI systems count as high risk?
Those that materially affect people's lives: CV screening and employee evaluation, credit and insurance decisions, admission and grading in education, safety components, and biometric identification. If you use one, you need assessments, documentation and demonstrable human oversight.
When do the AI Act obligations actually bite?
In stages: the prohibitions and the AI literacy duty from February 2025, the rules for general-purpose models from August 2025, and the bulk of the high-risk requirements from August 2026, with some extensions into 2027. Preparation has to start well before those dates rather than on them.
Let’s talk about your project
Message us on WhatsApp or send an email — you talk directly to a developer.
office@northdan.com · +40 752 070 247