Norway - AI Plan for Health Services

Overview

The Joint AI Plan for the health and care services was published by the Norwegian Directorate of Health (Helsedirektoratet) with a public release date of 18 February 2025. The plan is a coordinated, time‑bound sector framework covering measures for 2024–2025. It implements a mandate from the Ministry of Health and Care Services to facilitate the safe introduction and broader, appropriate use of artificial intelligence across primary and specialist health services in Norway. The Directorate led the work in collaboration with the Norwegian Medicines Agency (Medical Products Agency), the Norwegian Board of Health Supervision, the Norwegian Institute of Public Health, the regional health authorities and KS (the Association of Local and Regional Authorities). The plan aims to enable benefits from AI while safeguarding patient safety, privacy and rights, and to provide practical support for national, regional and municipal actors involved in procurement, deployment and oversight of AI systems.

Definitions

The plan provides and uses sectoral definitions for AI and related concepts (see the chapter on definitions in the full report). It distinguishes types of applications, including clinical decision support, administrative automation, large language models and other machine learning systems, and clarifies how these categories overlap with regulated categories such as medical devices. Definitions are aligned with common EU and Nordic terminology and with sectoral definitions used by contributing national agencies. Where necessary, the plan refers stakeholders to the detailed definitions section of the full publication for precise wording and sectoral context.

Governance and Institutional Framework

The Directorate of Health is identified as the lead coordinating body for the joint plan, with specified collaboration partners across regional authorities, inspectorates and sector organisations. Governance measures established by the plan include the formation of an AI Advisory Board for the health and care sector (expected to meet four to six times annually), structured cross‑agency collaboration among named agencies, use of a prioritisation committee to sequence activities, and establishment of user panels to incorporate provider and patient perspectives. The plan describes roles and responsibilities for national agencies, regional health authorities and municipal providers, and asks that agencies clarify responsibilities that rest with vendors, national bodies and local providers. The governance model is intended to support a shared approach to quality assurance, validation, guidance production and competence building, while ensuring accountability through existing supervisory authorities.

Key Focus Areas

• Sector coordination and cross‑agency collaboration: establishing mechanisms (AI Advisory Board, prioritisation committee, user panels) to coordinate activity across national agencies, regional authorities and municipal providers.
• Quality assurance and validation: developing a framework for quality assurance and producing a report on validation of AI solutions to support safe deployment.
• Regulatory guidance and navigation: producing cross‑agency regulatory guidance and public information pages to help services navigate existing legal frameworks (medical device rules, health legislation and data protection).
• Competence building and training: emphasis on building competence among health personnel and managers, including seminars and knowledge resources to elevate sector capability to assess, implement and monitor AI.
• Large language models and specialised resources: creating knowledge resources on large language models (LLMs) and preparing risk assessments specific to LLMs.
• Information sharing and lessons learned: running a series of seminars and building a knowledge base to share experience, use cases and sector needs.
• Benefit realisation and ethical use: promoting medically appropriate, ethically sound use that secures patient safety, transparency and respect for rights.
• European and Nordic cooperation: aligning with European and Nordic standardisation efforts, and preparing providers for EU regulatory developments such as the EU AI Act.

Implementation Framework

The plan sets out concrete measures and a programme for 2024–2025. Implementation responsibilities are split between the Directorate of Health as the lead coordinator and partner agencies that will produce guidance, validation tools and public information. Key implementation activities include establishing the AI Advisory Board, convening seminars and user panels to collect and disseminate experience, producing cross‑agency regulatory guidance and public information pages, developing a sector framework for quality assurance, publishing a report on validation of AI solutions, creating knowledge resources on large language models, and preparing targeted risk assessments and reports for the Ministry of Health and Care Services. The plan indicates that measures are intended to support local and regional implementation while clarifying national responsibilities for guidance and shared activities. It positions sectoral actors — national agencies, regional health authorities, municipal health and care providers, suppliers, and R&D environments — as primary implementers, with the Directorate facilitating coordination and consolidation of resources.

Monitoring and Evaluation

Monitoring and evaluation within the plan rely on a mix of cross‑agency reporting, milestone deliverables and existing supervisory channels. The plan documents specific milestone deliverables (for example, a knowledge base on large language models, a report on quality assurance and assessments to the Ministry) and asks agencies to report progress to the Directorate and to the Ministry where appropriate. User panels and the prioritisation committee are intended to provide ongoing feedback and to inform prioritisation of activities. The plan also links expected implementation and validation practices to the supervisory roles of existing agencies so that monitoring of compliance and safety is exercised through existing enforcement mechanisms. Regular seminars and knowledge‑sharing events are included as practical evaluation forums to share experience and surface implementation challenges.

Penalties, Liability, and Appeals

The Joint AI Plan is a coordinating policy and guidance document rather than primary legislation; it does not itself create new statutory penalties. Enforcement and penalties remain grounded in existing sector laws and the mandates of supervisory agencies. The plan explicitly links expected implementation and validation practices to the enforcement and supervisory roles of agencies such as the Norwegian Board of Health Supervision and the Norwegian Medicines Agency. Where AI applications fall within regulated categories (for example certain medical devices), existing rules, conformity assessment procedures and enforcement pathways apply. The plan seeks to reduce uncertainty by clarifying which agency’s rules are likely to apply and by providing guidance to help providers meet existing statutory obligations (patient safety, duties of health personnel, data protection and medical device requirements). Appeals and liability therefore follow the ordinary legal channels associated with the underlying statutory frameworks and regulatory bodies; the plan does not establish separate appeal mechanisms.

Relationship to Other Instruments

The plan positions itself as sectoral implementation support that complements national laws and EU developments. It explicitly seeks alignment with the EU AI Act and wider digital policies, and coordinates with national digitalisation strategies and existing health regulations (medical device rules, patient and data protection laws). The plan sets out how sectoral guidance and quality assurance work will interface with statutory obligations and with the responsibilities of agencies such as DigDir (the Norwegian Digitalisation Agency) and the Norwegian Data Protection Authority where relevant. The plan is intended to prepare providers for future regulatory developments by clarifying expectations, producing guidance for navigation of current rules and by aligning terminology and approaches with European and Nordic standardisation work.

International Alignment

The plan refers to and aligns with common EU and Nordic terminology and with international standardisation work relevant to AI in health. It seeks cooperation with European initiatives, aligns its definitions and guidance with commonly accepted EU/Nordic terms, and explicitly aims to prepare the sector for the EU AI Act and related European policy and standardisation processes. The plan also positions Norway’s sector activity within Nordic collaboration avenues and European standardisation work to ensure interoperability of approaches and to leverage shared knowledge on risk assessment, validation and governance of AI in health.

Implementation Timeline

Sources and References