Netherlands · Public sector policy
Algoritmekader and Algoritmeregister (Netherlands)
The Dutch government's Algoritmekader sets out the rules and tools for responsible public sector algorithms, and the Algoritmeregister publishes them.
The Algoritmekader is the Dutch government's framework for the responsible use of algorithms and AI by public bodies. It brings together the legal requirements, including the EU AI Act, with recommended measures and tools across a lifecycle that runs from organisational readiness to decommissioning. Its companion, the Algoritmeregister at algoritmes.overheid.nl, is the public register where Dutch government organisations publish information about the algorithms they use, with a focus on impactful algorithms and high-risk AI systems.
What it is
The Algoritmekader, or algorithm framework, is the Dutch government’s guidance on the main laws and rules for the responsible use of government algorithms, with tips and tools for applying them. It is published as an open website, maintained in the Ministry of the Interior’s public repository, and version 2.5 was launched on 24 March 2026.
The Algoritmeregister van de Nederlandse overheid is its public counterpart. On it, Dutch government organisations publish information about the algorithms they use and explain how they work. The register focuses on impactful algorithms, including high-risk AI systems, and listed 1,592 algorithms when this page was reviewed.
Who it applies to
Both are written for Dutch public bodies, from ministries and agencies to local government. They do not apply to private companies directly.
They matter beyond government in two ways. Suppliers selling algorithms or AI to Dutch public bodies will meet the framework’s expectations in procurement, and the framework includes model procurement terms among its tools. And because the register is public, anyone can see which algorithms a government organisation has published and how it describes them.
What it covers
Requirements. A catalogue of the legal requirements that apply to government algorithms, including the EU AI Act, the GDPR and Dutch law, with an AI Act section that includes a timeline and a summary.
Recommended measures. For each requirement, measures a public body can take to meet it.
A lifecycle. Measures are organised by phase.
- Phase 0, organisational responsibilities, before any algorithm is used or developed.
- Phases 1 and 2, problem analysis and design.
- Phases 3 to 5, data exploration and preparation, development, and verification and validation.
- Phases 6 to 8, implementation, monitoring and management, and decommissioning.
Roles. Guidance is tagged for policy and advice staff, lawyers, developers and project leaders.
Tools. A library of instruments, including the Data Protection Impact Assessment, the Impact Assessment Mensenrechten en Algoritmes (IAMA), the AI Impact Assessment, an impact analysis for the AI Act, the ALTAI checklist, a guide to inventorying, identifying and classifying AI systems, a non-discrimination by design guide, an AI literacy self-assessment, and the Algoritmeregister itself.
Topics. Cross-cutting guidance on discrimination, privacy and personal data, and public procurement.
Key dates
- 24 March 2026. Algoritmekader version 2.5 launched.
- 2 August 2026. EU AI Act general application, including transparency duties for chatbots and generated content.
- 2 December 2027. Annex III high-risk obligations apply. Public bodies deploying high-risk AI must carry out a fundamental rights impact assessment and register their use in the EU database.
How it relates to other frameworks
The Algoritmekader sits on top of binding law. The EU AI Act and the GDPR apply to Dutch public bodies directly, and Article 40 of the Dutch GDPR Implementation Act sets when they may decide automatically without profiling. Who will supervise the AI Act in the Netherlands is set out in the draft implementation bill.
The framework’s lifecycle and measures map closely to international frameworks. Organisations that use ISO/IEC 42001 or the NIST AI RMF will recognise the same steps of inventory, impact assessment, validation, monitoring and decommissioning.
A note on this page
This is Aicura’s summary of the Algoritmekader and the Algoritmeregister. It is not legal advice. Read the framework and the register on their official sites, and take advice on how the law applies to your organisation.
Last reviewed 18 October 2026. This page explains the framework to support your governance work. It is not legal advice; read the official source for the authoritative text.
Questions
Is the Algoritmekader binding?
The framework itself is guidance. It collects requirements that are binding in their own right, such as the GDPR, the EU AI Act and Dutch administrative law, with recommended measures and tools that are mostly optional. The framework notes that a data protection impact assessment is mandatory where the GDPR requires it, while most other tools are not.
Do Dutch government bodies have to publish in the Algoritmeregister?
The register is how Dutch government organisations publish information about their algorithms, with a focus on impactful algorithms and high-risk AI systems, and it lists more than 1,500 of them. Separately, the EU AI Act requires public authorities deploying high-risk AI systems to register their use in the EU database.
What is the IAMA?
The Impact Assessment Mensenrechten en Algoritmes, a human rights and algorithms impact assessment developed for Dutch public bodies. The Algoritmekader lists it among its tools alongside the DPIA, the AI Impact Assessment and an impact analysis for the AI Act.
Start with the AI you run
Every framework begins with knowing which AI systems and agents you have. The AI register is free to start.
