In a municipality's management team, the town clerk asks whether AI literacy has been taken care of. The CISO points to the spring lunch lecture: 140 attendees, a presentation on Copilot, and the agreement that no personal data goes into free tools. Then it goes quiet, because nobody is sure that covers a youth care consultant whose conversations are written up by a transcription tool.
It depends on the role. Article 4 of the AI Act asks a municipality for measures towards a sufficient level of AI literacy among everyone who uses AI on its behalf. Sufficient for the front desk is not sufficient for social services, and one lecture for the whole town hall does not cover that difference. What article 4 asks in general is in AI literacy in 2026.
When does article 4 apply to a municipality?
As soon as employees use an AI system for work. The regulation is addressed to providers and deployers, and a municipality where people use Copilot, a chatbot or a transcription tool is a deployer. Whether the municipality procured those tools itself makes no difference. The text is in Regulation (EU) 2024/1689, article 4, with the definition of AI literacy in article 3.
"On their behalf" reaches beyond the payroll. A municipality works with contractors, a joint arrangement for social services, an external communications agency. Whoever uses an AI system on the municipality's behalf falls under the measures, appointment or not.
There is no training in article 4, no certificate and no number of hours. The European Commission confirmed in its questions and answers on AI literacy that a certificate is not required and that an organisation may record its own measures. Supervision has sat with national authorities since August 2026; how that is divided in the Netherlands is not yet settled.
Why does it differ per role in a municipality?
Because the work differs, and so does the data in it. A front-desk employee makes an email easier to read, a policy adviser has research done with sources, a youth care consultant handles notes about families, and the town clerk explains the decision per tool to the council.
| Role | What sufficient means | What you record |
|---|---|---|
| Front desk, contact centre | Which tools are allowed, that the citizen service number and case file contents go in no prompt, and that an answer is checked before it goes to a resident | The explanation this group received, and the decision per tool |
| Policy adviser, communications | That an answer with sources still means sources you open yourself, and that a resident's case is not an example for a prompt | The same, plus the condition on research tools |
| Youth care, social support, debt counselling | That case work belongs in no unapproved tool, even without a name, and which route does exist | The instruction for social services and who was briefed on it |
| IT, CISO, DPO | Which tools are in use, what their status is, how to assess a tool and how the decision reaches employees | The register, the assessment questions and the decisions |
| Board, town clerk, management team | That the municipality decides and not the vendor, which applications may be high risk, and who decided what | The policy, the decisions and their dates |
For most employees it sits in the first two rows, where the explanation is short. What a municipal employee does and does not put in a prompt, per task, is in ChatGPT and personal data in municipalities.
When does it become more than an explanation?
When AI is used in a decision about a resident. Annex III of the regulation lists, among other things, AI systems public authorities use to assess whether someone is eligible for benefits and public services. A chatbot that rewrites a letter does not fall under that. A system that pre-sorts applications may. Whether a specific application is high risk is an assessment per system and per use, and it is the municipality's, not the vendor's. How to make it is in when is an AI application high risk.
Whoever works with such a system needs to understand how it reaches an outcome, what it misses, and when to deviate from it. This is also where the Dutch government's algorithm register comes in: public bodies are asked to describe impactful algorithms there publicly, and the explanation given to employees should match that description. Everyday AI use with chatbots and transcription is not in it, and needs its own internal register.
What does that look like in a municipality?
After the management meeting, the municipality from the first paragraph finds out what is in use. Next to Copilot, Perplexity, ChatGPT and three specialised tools are running, among them a transcription tool two youth care teams have used for months for conversations with families. No decision was ever taken on that tool, and the lunch lecture did not cover it, because in spring nobody knew it existed.
That is no proof the lecture failed. It explained what was known in May; the work went on with tools nobody then recognised as AI. It does show what a lecture cannot do: be present when a consultant starts a recording.
The municipality then does three things. It decides on the transcription tool, after a look at the vendor, the processing location and the terms. It writes a shorter instruction for social services. And it makes sure that whoever opens an AI tool sees the decision at that moment, with the approved alternative alongside.
How do you demonstrate it?
With three kinds of evidence. The explanation per role and who received it shows that you told people what they need to know. The register and the decisions show what was decided. The figures on what happens at the moment of work show whether that decision reached the working day. Which of the three an auditor weighs most, and what you honestly call not substantiated, is in how do you demonstrate AI literacy.
That third kind of evidence is what an attendance list lacks. A notice that appears when someone opens a tool can be counted with what the person did: opened the alternative or continued anyway. A marking of a citizen service number in a prompt can be counted with its outcome: masked, removed, or sent anyway. Counts only, never who.
