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

AI in Norway's news media

43 verified, source-linked AI implementations across 14 news organizations, 2018–2025. Norway ranks 10th worldwide by documented cases.
Europe and Central Asia2018–2025
Cases
43
Organizations
14
Active since
2018
Controversies
3
Compared with the global pattern, Research & Investigation accounts for 21% of cases here against 10% worldwide; and Editorial Workflow & Writing Assistance accounts for 26% of cases here against 17% worldwide.

What AI is used for — and how that differs

Norwayworldwide sharepp = percentage-point difference
Editorial Workflow & Writing Assistance11+9pp
Research & Investigation9+11pp
Content Production & Automation5-3pp
Audience Experience & Personalisation5+5pp
Multimedia & Visual Production3+3pp
AI Ethics & Editorial Policy2+1pp
Talent Development & Upskilling2-2pp
Audience Intelligence & Analytics1+1pp
Data Platform & Infrastructure1+2pp

How far adoption has gone

1 – Access1
2 – Adoption30
3 – Proficiency11
4 – Ways of Working1
5 – Reorganizing0

Who is adopting

Funding: Own budget (41), Tech company (OpenAI) (1), Public broadcaster / state media (1)

Notable cases

2024Aftenposten

Published editorial AI guidelines ("Retningslinjer for KI i Aftenposten"), framed as a living document under continuous revision. Four main rules: label generative-AI use clearly and understandably; treat AI-created material like any other source — critically and curiously; the newsroom must quality-assure all AI use; and sensitive cases or information must not be shared with AI services before publication.…

AI Ethics & Editorial Policy2 – Adoption
Editorial AI policy (labelling, source-criticism, quality assurance, sensitive-data rules,Source →

2025Verdens Gang (VG)

VGX — AI-native news product reinventing VG's editorial format (launched in beta Dec. 2025 with 700 beta users); replaces traditional articles, front pages and pre-roll video with a single feed combining overview and in-depth layers; editorial content from Schibsted publications is broken into 'journalistic atoms' and reassembled by AI agents acting as reporters, editors and editors-in-chief; entire AI agent…

Audience Experience & Personalisation1 – Access
VGX (Schibsted/VG proprietary AI agents + journalistic atoms framework; article-to-video gSource →

2022iTromsø

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…

Research & Investigation2 – Adoption
Djinn — custom scrapers/APIs + BERT (Norwegian NLP) + Meta Llama (summarisation)Source →

2023Schibsted

WAAS (Whisper as a Service) — open-source GUI and API wrapping OpenAI's Whisper speech-recognition model, with job queuing, so a whole newsroom can self-host transcription instead of sending audio to a commercial service. Released by Schibsted under Apache 2.0 and maintained since 2023.

Transcription & Translation3 – Proficiency
WAAS — open-source Whisper GUI/API with queuing (github.com/schibsted/WAAS, Apache 2.0)Source →

Amedia

AI Sandbox — a secure, self-hosted generative-AI environment (launched Feb. 2024) that lets journalists across Norway's second-largest local-media group experiment with GenAI on the group's own paywalled content without exposing it to external LLMs. Paired with group-wide generative-AI guidelines under which local editors bear ultimate responsibility for AI use.

Data Platform & Infrastructure2 – Adoption
In-house generative-AI sandbox (self-hosted; keeps paywalled content out of external LLMs)Source →

2023Polaris Media / iTromsø

DJINN (Data Journalism Interface for Newsgathering and Notifications) scaled to 35 Polaris Media newsrooms — extracts documents from municipal archives, summarises and ranks them by newsworthiness using journalist-trained scoring; reduces document research from 2-3 hours to ~5 minutes per story; rollout began August 2023, reaching 35 newsrooms by February 2024; six cover stories generated in first week at…

Research & Investigation3 – Proficiency
DJINN (in-house, developed with IBM)Source →

Elsewhere in Europe and Central Asia

Albania 4Armenia 7Austria 22Azerbaijan 1Belarus 5Belgium 25Bosnia and Herzegovina 3Bulgaria 6Croatia 5Cyprus 1Czech Republic 24Denmark 27
Data from J·Index, the Journalism AI Index — every case verified against a published source. See the methodological note for how cases are collected and coded. Updated 2026-07-23.