The algorithm register: must governments publish their AI?

Short answer: Yes, in two ways. Dutch public bodies publish the algorithms they use in the national Algorithm Register (algoritmes.overheid.nl), as a transparency instrument towards citizens. In addition, the AI Act requires registration of high-risk AI in an EU database (Art. 49 and 71) — and authorities deploying such a system also register their use. Two registers with partly different aims, but the same core: accountability.

The national Algorithm Register

The Algorithm Register is a Dutch transparency instrument: per algorithm, public bodies describe what it does, which data it uses, how oversight is arranged and who is responsible. It has been made mandatory in phases for public organisations, starting with the most impactful (high-risk) uses. The aim: citizens can see which algorithms may affect them.

The EU database under the AI Act

The AI Act adds a European layer. Providers of high-risk AI must register their system in an EU database before placing it on the market (Art. 49). For public-sector uses, the deploying authority also registers the deployment (Art. 71). This creates EU-wide visibility of where high-risk AI is used.

How they connect

The national register serves broad public transparency; the EU database serves regulators and the internal market. They overlap for high-risk government AI: register in both, with consistent information. A good FRIA often already provides the content you need for a register entry.

What to do

• Inventory which algorithms/AI you deploy and classify them (see AI in government).
• Publish the impactful uses in the Algorithm Register.
• Register high-risk AI in the EU database (provider and, for public deployment, deployer).
• Keep the information consistent with your FRIA and technical documentation.

For government AI, transparency is not a final step but a precondition: registering makes traceable what an algorithm does — and that is exactly what public trust rests on.