Denmark has quietly deployed at least 129 AI systems across 51 public authorities, with no public register established to show where algorithms process citizens’ data, even as new legislation gives ministers power to restrict residents’ data rights when AI is involved.
The Danish Data Ethics Council recommended in 2023 that the government create a national public register of all generative AI systems used by public authorities. No such register has been established. According to Datatilsynet’s mapping of AI use in the public sector, roughly one quarter of the 191 surveyed authorities already use AI solutions, averaging 2.5 systems each. The scale is concentrated but largely invisible to the public.
Where Your Data Goes
AI is already embedded in Danish public administration. According to Datatilsynet’s mapping, municipalities account for 65% of AI-using authorities, with state bodies making up the remainder. These systems handle everything from invoicing and classification to citizen-facing chatbots. Digitaliseringsstyrelsen’s generative AI catalogue lists concrete AI projects at municipal level, including digital assistants that explicitly target foreign-language speakers who struggle with Danish self-service platforms.
For expats who interact with multiple authorities for tax, residence permits and municipal services, exposure to AI-mediated decisions is disproportionately high. There is no consolidated transparency tool or multilingual explanation of where AI enters the process. There is also no straightforward way to know whether an algorithm has flagged your case for review.
New Laws Expand State Power
Two legislative initiatives are transforming Denmark’s AI governance. L 111, the national supplement to the EU AI Act, establishes the supervisory framework for AI regulation. A separate February 2026 bill gives all public authorities a clear legal basis to use AI systems for development, operation, decision support and profiling based on personal data they already process.
The second bill includes a critical clause. According to legal analysis by Poul Schmith, Section 4 grants the competent minister power to adopt rules that restrict or suspend data subjects’ GDPR rights, including access and correction, when Article 23 conditions such as national security or crime prevention apply. This represents a significant expansion of ministerial discretion over data rights in an AI context.
The bill prohibits purely automated decisions without human involvement. However, it does not define what meaningful human review actually requires.
High-Risk AI Systems: A Staggered Timeline
The EU AI Act’s obligations are phased in gradually. Rules on forbidden AI practices took effect on 2 February 2025, obligations for general-purpose AI models from 2 August 2025, and high-risk system requirements covering areas such as migration, welfare and policing apply from 2 August 2026, with some product-linked systems following in 2027. Denmark’s national legislation is being finalized precisely as these obligations take hold.
Efficiency Over Transparency
According to the Confederation of Danish Employers (DA), 37,000 administrative full-time equivalents can be eliminated from the public sector by 2040 using AI. That breakdown includes 18,000 from the state, 13,000 from municipalities and 6,000 from regions. DA also estimates a 12% productivity lift across the sector and calls for binding targets on headcount reduction.
DI Digital, the tech industry lobby, argues that bureaucracy is blocking AI projects that could free staff time and improve services. The government’s coalition programme sets a goal of freeing at least 30,000 full-time equivalents and deploying a national digital health assistant by 2028. The focus remains on shrinking the administrative machine, rather than on transparency or citizen control.
Where the Ethics Council Stands
The Data Ethics Council urges public authorities to begin with low-risk generative AI use cases, systematically assess risk profiles and scale human oversight accordingly. As reported by the Council in its generative AI report, centralized guidelines on error handling and clear communication on quality control are also priorities. The call for a public AI register is explicit, but the government has not acted on it.
Datatilsynet’s mapping shows many authorities lack clarity on legal bases and risk management for AI solutions. Governance weaknesses are visible, but enforcement remains patchy.
What You Can Do
In practice, there is limited scope to avoid AI use by Danish public authorities when filing cases digitally. Under GDPR, however, any resident including non-citizens retains rights to access information about how personal data is processed, to request correction and to limit processing in some circumstances. The AI-use bill allows those rights to be curtailed where ministers invoke Article 23 conditions, so full transparency cannot be assumed in high-risk domains such as migration or policing.
Datatilsynet remains the primary body for complaints related to data protection and AI use. Digitaliseringsstyrelsen’s generative AI catalogue lists concrete AI projects and can help identify where AI might touch interactions with the state. The Data Ethics Council’s reports are publicly available and can be used by journalists and NGOs to press for action on the missing register.
The Bigger Picture
Denmark has a well-documented history of early digitalisation, including mandatory digital post and MitID, which previously created barriers for non-Danish speakers and newly arrived residents. AI is the next phase of that trajectory, and it is being rolled out with a similar combination of efficiency ambition and transparency gaps.
As confirmed by L 111’s explanatory notes, the EU AI Act is an internal-market instrument designed to ensure the free movement of AI-based products and services, which limits how far any Member State can diverge from its framework. DA’s explicit 37,000 FTE reduction target remains unusually quantified by European standards, even if direct comparisons are difficult without a full EU-wide survey.
For expats who already navigate a system built around Danish speakers, Danish ID numbers and Danish bank accounts, AI adds another layer of opacity that oversight frameworks have not yet caught up with.
