Generative Artificial Intelligence in Norway

The report is available in Norwegian only.

Artificial intelligence (AI) refers to machines that can perform both physical and cognitive tasks previously reserved for humans. The technology is becoming more powerful and versatile, but at the same time, it is becoming harder to understand and control.

Generative artificial intelligence is already contributing to scientific discoveries and innovative solutions to complex problems in several sectors.

While information technology has so far mainly affected manual and rule-based work, generative artificial intelligence will impact knowledge workers and creative professions. The technology can replace certain jobs and automate tasks, but it can also complement employees’ expertise.

The technology is resource-intensive to develop and lacks transparency. This has so far resulted in a concentration of knowledge and power to define how the technology is designed and functions within the largest tech companies.

With generative artificial intelligence, it has become much easier to create artificial yet convincing audio clips, images, and videos. This opens the door to more advanced forms of fraud, abuse, and disinformation.

Opportunities for Norway

The EU’s AI regulation provides Norway with a solid foundation to regulate artificial intelligence and offers a new and strengthened opportunity to develop national policies. Norway must act swiftly and strategically to take advantage of this opportunity and ensure that the technology is used in a safe, responsible, fair, and beneficial manner for society.

Expert Group Recommendations

1. Distribute oversight among several bodies, with one strong coordinator. Norwegian authorities should allow sector-specific regulators to enforce the EU’s AI regulation in their respective areas, while appointing a central coordinator equipped with AI expertise and technical resources to assist sector regulators as needed.
2. Establish a national AI incubator. The AI regulation requires regulatory sandboxes for artificial intelligence. Norwegian authorities should create a new national sandbox under the Directorate for Digitalisation, with an expanded mandate as a national AI incubator for public administration.
3. Provide public administration with a problem-solver. There is a need for continuous and collective handling of uncertainties and issues related to generative artificial intelligence within public entities. A network should be established under the leadership forum “Skate” (Steering and Coordination of Services in E-Government).
4. Ensure access to high-quality Norwegian data. Large amounts of high-quality Norwegian data are required for language models to function well in Norwegian. The National Library should be given the mandate to make such data available and compensate the rights holders. Norwegian data must be strategically managed, and access should not be granted indiscriminately.
5. Invest in computing power as national infrastructure. The need for computing power will increase in the coming years, and more powerful supercomputers must be built in Norway. Computing power should be prioritised as a national infrastructure investment.
6. Offer Norwegian language models as a national shared service. Access to Norwegian-developed language models will be essential for developing and delivering competitive digital services in the future. These models should be developed and operated as a national shared service and infrastructure.
7. Set quality standards for Norwegian language models. There will soon be many language models with different capabilities. A designated research institution should be tasked with developing quality and evaluation criteria for language models in Norway, providing a foundation for informed choices of models.
8. Establish rules for labelling AI-generated content. False and artificially generated content that appears real is spreading quickly online. Norwegian authorities should develop guidelines for transparency and labelling of AI-generated content.
9. Create rules for generative artificial intelligence in election campaigns. Generative AI is increasingly being used as a campaign tool and to influence democratic elections. Before the 2025 parliamentary election in Norway, cross-party rules for the use of generative artificial intelligence in election campaigns should be developed.
10. Scale up efforts for fact-checking. The rise of false, AI-generated content increases the need for fact-checking in public discourse. Efforts should be scaled up by strengthening Faktisk.no (a non-profit organisation dedicated to fact-checking public debates and discourse in Norway) or establishing an independent national centre for source verification. Investment should also be made in technical tools and methods for content verification in Norwegian.
11. Create a psychological defence. The rise of generative artificial intelligence has increased the spread of disinformation and information manipulation. A national function should be established to identify and advise on disinformation campaigns and information manipulation by foreign actors.
12. Strengthen efforts on AI security. Generative artificial intelligence introduces new security risks and has great potential for harm. National capacity to analyse threats and develop risk scenarios should be formalised and strengthened. Authorities should also develop Norwegian positions on global governance mechanisms for artificial intelligence and ensure that Norway has a seat in the new governance bodies at the EU level.