Challenges
The most capable AI systems are built by a few companies in two countries, and they are quickly becoming part of how public services, businesses and citizens work. In 2026, export controls cut access to leading models, China drafted its own export restrictions on AI, and AI safeguards blocked defenders during cyber attacks. At the same time, AI agents are starting to act on people’s behalf, and public services were not built for them. Failure is not always sudden. A service can stay online while its quality, accountability or the skills behind it slowly decline. The region also has few suppliers of its own in key parts of the AI stack, such as general-purpose models, computing power and tooling.

Project purpose
The paper aims to give decision-makers a shared understanding of the region’s AI dependencies and the choices available. It sets out the evidence and describes options rather than prescribing a single strategy. It asks one central question: for each important service, how much does it need the best AI, and how ready must it be if that AI fails, changes or disappears? It also explores where the eight countries could achieve more together than alone.

Expected outcomes

  • A common definition of AI resilience with four parts: continuity, adaptability, assurance and accountability, and capability
  • A first picture of the region’s AI supplier landscape, drawing on the full-stack mapping
  • A set of options for governments, public bodies and businesses, such as exit terms in contracts, tested fallbacks, shared testing of AI systems and joint purchasing
  • Candidate areas for regional cooperation, including shared public-sector pilots, shared testing in the region’s languages and shared reporting of AI incidents
  • Seven open questions to guide discussion at the High-Level Forum and the regional work that follows

Funding
The white paper is funded by the Nordic Council of Ministers through New Nordics AI’s core funding.

The paper launches at the New Nordics High-Level Forum in Copenhagen on 6 November 2026, together with the New Nordics AI full-stack mapping of the region’s AI companies. The full-stack mapping was prepared by August Associates with company data from Dragonfly, and it gives the paper its evidence on where the region stands.

‍

‍