← J·Index — The Journalism AI Index
J·Index / Norway / iTromsø

iTromsø

2 verified, source-linked AI implementations documented between 2022–2025.

NorwayEuropeLocal / Independent
Cases
2
Active since
2022
Peak stage
2
Controversies
0

Where the AI sits

Editorial Production2

AI implementations

2025

How iTromsø Built Norway's Most AI-Ready Newsroom

Research & Investigation2 – Adoption3 – Collaborative Human–AI
Tool: CustomSource ↗
2022

Djinn

An AI-powered data journalism interface built by iTromsø, a 20-person northern-Norwegian newsroom in the Polaris Media group. Custom scrapers and APIs collect over 12,000 municipal PDF documents a month; machine learning ranks them by relevance, auto-summarises and extracts key information, and pattern detection surfaces investigative leads (key individuals, hidden connections). Built in about two months (~1,000 hours) by a ~15-person team with IBM and local firm VC2, using BERT for Norwegian and Meta's Llama for cost-effective summarisation.

Research & Investigation2 – Adoption3 – Collaborative Human–AI
Tool: Djinn — custom scrapers/APIs + BERT (Norwegian NLP) + Meta Llama (summarisation)Source ↗

Context

iTromsø is one of 15 organizations in Norway documented in the index, which holds 41 cases from the country. Its work spans 1 function type, including Research & Investigation, documented across 131 organizations worldwide.

Explore related

More newsrooms in NorwayResearch & InvestigationHow this is verified
Part of J·Index — The Journalism AI Index. Every case carries a live source link. Cite with attribution; bulk reproduction is not licensed. Data licensed CC BY-NC 4.0.