Denmark - National AI Taskforce

Overview

The Digital Taskforce for Artificial Intelligence is a cross-public-sector framework established by the Danish Government together with KL and Danske Regioner to accelerate responsible, large-scale deployment of AI across the public sector. The taskforce was announced as part of Denmark's national strategic initiative for AI on 2 December 2024 and set out in its formal Kommissorium. Its mission is to identify and remove legal, organisational and technical barriers (explicitly referencing GDPR and the EU AI Act), publish a target vision (målbillede) for AI in the public sector, run and scale pilots and testbeds, and ensure that public deployments are safe, transparent and rights-respecting. The taskforce is anchored in the Digitalisation Ministry and hosted by a multi-ministerial secretariat to enable coordination across policy areas and sectors.

Definitions

Key definitions used in the framework rely on the national strategic publication and Kommissorium. 'Artificial intelligence' and 'generative AI' are defined broadly to include systems that perform or assist with tasks traditionally requiring human intelligence (pattern recognition, decision support, natural language generation). 'High-risk' follows EU AI Act terminology where relevant for public sector uses that affect legal status, significant interests or sensitive rights. 'Public-sector solution' refers to AI tools procured, developed or used by ministries, agencies, regions and municipalities. The Kommissorium also distinguishes 'pilots' (time-limited evaluations) from 'scale deployments' (integrated production services), and emphasises the need for data quality, explainability and traceability in documentation and lifecycle governance.

Governance and Institutional Framework

The taskforce is organised with a secretariat anchored in the Digitalisation Ministry, drawing staff from the Finance Ministry, Justice, Health, Climate and other relevant ministries as well as representatives from KL and Danske Regioner. A board composed of department heads provides strategic direction, while a political steering committee (finance minister, digitalisation minister and the chairs of KL and Danske Regioner) oversees progress and resource allocation. The Kommissorium specifies half-yearly reporting to the political committee and routine status updates to government. An external follow-up/expert group composed of researchers, industry and practitioner representatives provides technical and ethical advice; further external procurement and specialist contracts may be used for technical evaluations. The institutional model is therefore hybrid: governmental ownership for policy and funding decisions; cross-sectoral operational delivery by the secretariat; and independent technical and ethical input from the expert group and partners. The governance design deliberately aligns with sector responsibility principles (sectoral implementation by the responsible authority) while offering central coordination and removal of cross-cutting barriers such as interoperable data access and regulatory interpretation (Kommissorium).

Key Focus Areas

The taskforce concentrates on several interlocking focus areas to enable responsible scaling of AI in public services. First, strategic target-setting: publishing an initial målbillede with metrics and sectoral prioritisation to set ambitions for resource gains and service improvements. Second, legal and regulatory alignment: identifying where GDPR, the EU AI Act, the Medical Device Regulation and other sectoral rules pose implementation barriers and proposing clarifications or legislative changes. Third, operational pilots and scaling: coordinating cross-municipal and regional pilots, testing vendor solutions, and validating technical and organisational readiness for production. Fourth, data governance and access: improving availability and quality of datasets for safe model development (including initiatives for curated Danish text corpora and secure platforms for national language models set out in the December 2024 strategy). Fifth, safety, testing and evaluation: developing templates and methods for risk assessment, validation, continuous monitoring and model documentation (including obligations for transparency and record-keeping). Sixth, capacity and skills: supporting training, knowledge exchange and competence building across the public sector. Finally, public trust and ethics: ensuring deployments uphold citizens' fundamental rights, transparency and explainability, with targeted public communication and stakeholder engagement to mitigate exclusion or bias risks (Strategic publication, Dec 2024).

Implementation Framework

Implementation is operationalised via a phased approach: (1) produce the målbillede and sector prioritisation (early deliverable), (2) select priority pilots and establish testbeds with clear success metrics and legal assessments (DPIAs where personal data is involved), (3) run iterative evaluations and adjust legal/organisational barriers, (4) scale successful pilots to production across municipalities and regions, and (5) institutionalise procurement, compliance and ongoing monitoring practices. The Kommissorium mandates the secretariat to coordinate technical, legal and organisational resources, and to ensure that proposed solutions are implementable in a secure and rights-respecting manner, including where necessary engagement with the Danish Data Protection Agency (Datatilsynet) and other supervisory bodies. Funding is to be secured through ongoing budgetary agreements (finance minister and partners), and the initial operational timeframe is provisional until the end of 2027 with a mid-2027 decision point for extension.

Monitoring and Evaluation

The taskforce reports at least semi-annually to the political steering committee and provides ad hoc reports to ministers. Monitoring uses quantitative and qualitative indicators from pilots (e.g., time savings, FTEs released, service quality metrics, error rates, fairness/ bias measurements) and legal-compliance checklists (e.g., DPIA completion, contractual safeguards, vendor security certifications). Independent evaluations and periodic external audits are foreseen to validate claims and surface unintended consequences. The framework emphasises iterative learning: lessons from signaturprojekter and pilots (including earlier projects reviewed by Deloitte and others) feed into continuous refinement of templates, procurement guidance, data access arrangements and risk management playbooks.

Penalties, Liability, and Appeals

The Digital Taskforce itself is a coordination and enabling framework and does not introduce novel criminal or administrative sanctions. Liability and penalties for non-compliance with applicable law (e.g., GDPR breaches, procurement rules, sectoral safety rules) remain with the responsible public entity and are enforced by the competent authorities (e.g., Datatilsynet for data protection matters). The Kommissorium states the taskforce may recommend legal or regulatory amendments where current rules obstruct responsible deployment; any new punitive regimes would require legislative change. Appeals related to administrative decisions remain governed by existing Danish administrative law and sector-specific appeal routes.

Relationship to Other Instruments

The taskforce explicitly operates in the regulatory context of the EU AI Act, the GDPR, the Medical Device Regulation (where medical AI is concerned), procurement law and Danish administrative rules. It complements and implements the national Strategic Initiative for AI (Dec 2024) and will coordinate with sectoral agencies (Health, Social Services, Tax, etc.) to ensure sector responsibility. The Kommissorium instructs taskforce activities to consider interplay with ongoing EU-level work, EDPB guidance, and national data-ethics recommendations. It therefore serves as an implementation bridge between high-level strategy, EU-wide legal obligations and municipality/region-level operational practice.

International Alignment

The taskforce encourages international cooperation and learning, allowing partnerships with other leading countries to pilot and evaluate AI solutions. It aligns national ambitions with EU-level regulatory developments (notably the AI Act) and seeks to ensure Danish deployments maintain compatibility with cross-border data flows, research collaboration and common certification schemes. The December 2024 strategic publication highlights the need to combine national competitiveness (supporting domestic model development, Danish language models and open data) with compliance to international standards and norms to sustain both innovation and interoperability with EU partners.

Implementation Timeline

Sources and References