AP Consults on Explaining Automated Decision-Making Tools

AP asks for input on explanations in automated decision-making

21 April 2026 Themes: Algorithms explained Algorithms, AI and the GDPR The Autoriteit Persoonsgegevens (AP), the Dutch data protection authority, develops tools for how organisations should explain automated decision-making. In order to ensure that these tools fit in well with practice, the AP asks organisations, experts and stakeholders for input through a consultation.

Organisations are increasingly switching to automated decision-making to make certain decisions. Consider the assessment of a credit application or online applications. Organizations often use algorithms and artificial intelligence (AI) for automated decision-making.

Right to an explanation of an automated decision

If an organisation automatically makes a decision about someone, and that decision has serious consequences for them, that person is entitled to an explanation of this decision. So that they can understand the reasons behind the decision and, if necessary, defend themselves against the decision.

The mandatory explanation also forces organisations to understand how their own decision-making processes work.

What an explanation contains

In the explanation, the organisation must state:

• whether automated decision-making was used for the decision;
• the expected impact and importance of the decision;
• the underlying logic of the decision;
• what rights the person in question has and how they can exercise those rights. The explanation should not be vague or complicated. The explanation must also be complete.

Organisations that make automated decisions should provide general information about the decision-making process. They must provide that general information to all data subjects (the people whose personal data they process). The organisation must also provide personal explanations if requested by the person about whom an automatic decision has been made.

Thinking about explanations in advance is essential

Explaining can be complicated. For example, when the system that made the decision is not immediately transparent. When organisations work with algorithms that can't be explained, they can't tell the individual why a particular decision was made. That is why it is essential to think about explanations when choosing the system and the layout.

Some systems are in itself transparent to people. Other systems that need additional techniques to make them transparent. In addition, there are systems that cannot (yet) be explained properly.

Tools give practical help

The AP provides examples and practical help to organisations to make their explanations understandable. The AP also discusses what an organization should do if other interests would be affected by the explanation, such as the right to data protection of another person or a trade secret.

The tools are an extensive reference work for those who need to arrange explanations within an organization, and for people who receive an explanation and want to know what they are entitled to.

When publishing the final guidelines, the AP also publishes a roadmap for organisations and an overview for data subjects. Together with researchers from Utrecht University, the AP is also researching the effectiveness of different forms of communication in explaining algorithmic decision-making.

Your input is welcome

The AP invites organisations, experts and stakeholders to respond to the consultation. We look forward to insights from practice. For example, do you have experience with an approach that works? Or are you running into difficulties?

We use the input to refine the guidance. We will publish a summary of the responses received, without names of individuals and organisations and without contact details.

Would you like to respond? Please send an e-mail to ppa@ remove-this-text. autoriteitpersoonsgegevens.nl. You can also contact us with questions about the document or the consultation process. We would like to receive your input by 26 May 2026.

Publications

Consultation explanations for automated decision-making

PDF, 811 KB