Norway has decided to take artificial intelligence in healthcare seriously – safely, coordinated, and over time. Here's what you need to know, and what it means for your everyday life.

Artificial intelligence in healthcare has moved from being a future concept to something many already encounter in their jobs. An AI search for health information, a pilot in a department, a new tool that writes draft patient records. But as the technology has advanced rapidly, questions have followed: Is it safe? Who is responsible? How do we know it's actually good for the patient?

These are the questions the national initiative on AI in healthcare aims to answer. The Directorate of Health is now continuing the work with a joint AI plan for the health and care services for the period 2026–2027, and the direction is clear: artificial intelligence shall be used – but safely, coordinated, and based on knowledge. For you who document patient conversations every day, this is more relevant than it may sound.

What is the common AI plan?

The joint AI plan for health and care services for 2026–2027 carries forward the working methods and direction from the 2024–2025 plan. The overarching goal is to increase the use of safe and effective AI systems in healthcare and social services – solutions that should contribute to services of equal or better quality, and that free up time for healthcare personnel. The plan is built around three focus areas: sector collaboration on AI, frameworks and guidance, and a public AI service for targeted health advice.

Work does not happen in a vacuum. The Directorate of Health coordinates efforts with, among others, the Norwegian Medicines Agency, the Norwegian Board of Health Supervision, the Norwegian Institute of Public Health, the four regional health authorities, and KS. the joint AI plan was first launched in 2024, an intentional agile approach was emphasized: measures are prioritized and initiated continuously based on needs, instead of being locked into one large plan. A central task is to facilitate language models adapted to Norwegian conditions.

It's worth noting what the plan not It is not a mandate for everyone to start using AI tomorrow. It is a framework that will make it safer to adopt good tools when they are ready – with risk assessment, validation, and allocation of responsibility in place.

A safe introduction takes time – and that's a strength.

One of the most important realizations in the Norwegian work is that the safe introduction of AI takes time. The Directorate of Health has made initial assessments for a Public AI service for health-related questions to the citizens, and the path forward to a fully-fledged service requires thorough validation and quality assurance of the knowledge sources. It's not just about technology, but about trust.

At the same time, concrete expertise is being built within the service. Vestre Viken is leading a national research project – «SMART journal» – which, with 25 million kroner from Pilot Helse, will develop and validate AI assistants integrated into the electronic health record. A stated goal is to develop methods and standards for the safe validation of AI models in clinical practice, which is currently lacking. The project involves both somatic and mental healthcare, and researchers are looking at how AI can best understand clinical context, among other things.

It is this combination – ambition and caution – that characterizes the Norwegian approach. We have previously described how Norwegian e-health is moving from pilot to everyday use, and the AI plan is a natural extension of just that development.

What does this mean for you as a documentarian?

For the individual clinician, it's easy to perceive national plans as something distant. However, the direction impacts everyday work quite directly. When authorities emphasize language models adapted to Norwegian conditions, secure validation, and that AI should free up time for healthcare personnel, they are in practice describing the requirements a good documentation tool should already meet today: that it understands Norwegian medical language, that responsibility remains with the clinician, and that privacy is real.

It's worth remembering that no matter how much technology develops, legal responsibility remains fixed. The clinician is still the owner of the record and makes the final assessment. AI can write a draft, but the human approves. The national plan reinforces this principle rather than weakening it.

This is how Medivox fits into the picture

Medivox is built for the everyday life the AI plan points towards. The tool listens to the consultation and delivers a structured draft of the medical record before the next patient arrives – in Norwegian, adapted to medical language. Personal data is pseudonymized before the data is processed further, and all data processing takes place in Norwegian data centers. You build your own templates, and you own the medical record and make the final assessment.

There is no contradiction between a cautious national introduction and the fact that good tools are already in use. On the contrary: when the framework for safe AI is strengthened, it becomes easier for healthcare personnel to choose solutions that are already built around the same principles. We have previously written about how the clinic can prepare regardless of which way the great debate goes – and that advice also holds true here: You don't need to wait for everything to be fully investigated before you start saving time on documentation, as long as the tool you choose prioritizes safety and professional responsibility.

Frequently Asked Questions

What is the common AI plan for health?
This is the Norwegian Directorate of Health's coordinated plan for the safe and effective use of artificial intelligence in health and care services, continued for the period 2026–2027. The goal is increased use of safe AI systems that free up time for healthcare personnel.

Does the plan mean I have to use AI?
No. The plan is a framework for safe introduction, not a mandate. It is intended to make it safer to adopt good tools when they are ready.

Who is responsible when AI writes the medical record?
The clinician. AI can create a draft, but it is the healthcare professional who reviews, corrects, and approves it. You own the record and make the final assessment.

What is a SMART journal?
A national research project led by Vestre Viken, funded with 25 million kroner from Pilot Helse, which will develop and validate AI assistants integrated into the electronic patient record.

Do voice-to-text tools like Medivox meet the requirements of the plan?
Medivox is built around the same principles the plan emphasizes: Norwegian medical language, pseudonymization before further processing, data processing in Norwegian data centers, and the professional responsibility remaining with the clinician.

Use Medivox for free – Get started completely free

Do you want to understand how a secure, Norwegian documentation tool works in practice? Contact us – so we will go through how Medivox safeguards both privacy and professional responsibility.

Sources: