What is the AI Act?
The AI Act is the EU regulation on artificial intelligence. It sorts uses of AI by risk: some are banned, some are high-risk with strict requirements, some come with transparency duties, and most have no special requirements at all.
Does it apply now?
Partly. The bans have applied in the EU since February 2025 and the transparency rules since 2 August 2026. The high-risk requirements were moved to 2 December 2027 for most systems, and to 2 August 2028 for AI built into regulated products.
Norway is outside the EU but inside the EEA. As of October 2026 the AI Act is not yet Norwegian law. The government plans a new consultation in autumn 2026 and aims to put the law before parliament in spring 2027. If you sell AI products to customers in the EU, the EU rules can already apply to that part.
What does a chatbot have to do?
People must know they are talking to a machine. In the EU, since 2 August 2026, a chatbot has to make clear it is AI, at the latest at the first interaction, unless that is already obvious.
AI-generated content must also be labelled: images, audio and video made or altered by AI, and text published to inform the public.
It is worth doing anyway. It is simple, and it builds trust.
What is high-risk AI?
High-risk AI is AI used for decisions that can affect people's lives and rights, for example in recruitment, credit scoring, education and exams, access to public services and critical infrastructure.
High-risk systems need, among other things, a risk assessment, documentation, a record of what the system does, and a person who can follow along and step in.
What is banned?
Some uses have been banned in the EU since February 2025. They include manipulation that exploits people's vulnerabilities, social scoring of people, emotion recognition in workplaces and schools, and scraping faces from the internet at scale to build databases.
What should we do now?
Start with a simple overview of where you use AI and what for. Most companies find they have little or nothing that is high-risk, but a few places where people should be told they are talking to a machine.
Make sure chatbots and AI assistants say that they are AI. Label AI-generated images and text where it makes sense.
Check whether any of your use could count as high-risk, such as AI that sorts job applicants. If so, start documenting now.
Let a person make the decisions that matter, and make it possible to see what the AI did and why.
How I can help.
I build AI agents and chatbots designed to follow the rules from the start: they say what they are, show where an answer comes from, write down what they do and let a person take over.
If you already have a chatbot or an AI tool, I can go through it and tell you what needs to be in place, in plain language.