← ALL NEWS

VERSION2 · 13 Nov 2026 · DENMARK

Editorial: From hospital records to industrial secrets, Denmark is failing independent data processing

This editorial from Ingeniøren, written by editor-in-chief Henning Mølsted, argues that Denmark is too naive about who processes its critical data, putting citizens' privacy and Danish industry and research at risk of spying and leaks.

It points to two recent cases. Ingeniørens ComplianceTech found that sensitive patient data, such as information on anxiety attacks and diagnoses, was sent for AI processing in a tool built by the US health-IT supplier Epic and run on a Microsoft cloud platform. Weekendavisen reported on an enzyme for medical use discovered by Anthropic with AI. There is a suspicion that part of the breakthrough rests on University of Copenhagen research, which had used Anthropic's AI tool for coding and manuscript editing, and a suspicion of plagiarism has been raised.

The editorial notes that Aula runs on Amazon cloud services and that municipalities buy Microsoft Azure OpenAI assistants. Datatilsynet has warned Helsingør and other municipalities that Google products for schools risk children's data being used commercially. Meanwhile Denmark is getting three big Microsoft data centres, a Google data centre and possibly a Meta one, without a clear strategy for independent platforms. The author contrasts this with transport, where the state owns most roads, rail and airports, while data centres, cloud infrastructure and cables are privately owned, which Magtudredning 2.0 called "infrastructural power". A central concern is the US Cloud Act, which lets US authorities demand data held by Microsoft, Amazon and Google even when it sits in European data centres.

The editorial calls for public, perhaps joint European, data processing for the most sensitive data from industry, the state and universities, so that firms like Novo, Terma and Vestas and municipalities have options. It cites positive steps: the Danish Defence Intelligence Service is building a data centre in Copenhagen, and the Defence AI Center uses the Gefion supercomputer, which the state and the Novo Nordisk Foundation own three quarters of. Open-source models on closed servers and European clouds show a practical middle way.

It acknowledges the downsides: dropping leading models from OpenAI, Google or Anthropic could slow Danish productivity and innovation, and matching Big Tech's security against Russian or Chinese threats would take very large investment.

Read the original ↗