A form tutor at a secondary school asks a chatbot to soften the tone of a pupil's report comment. The pupil's name and the reason for concern are still in it. One floor up, the special needs coordinator asks the same tool to summarise a support plan for tomorrow's meeting. Last year the board, which runs nine schools, adopted an AI policy that mentions neither situation.
AI governance in education is about that gap. It is the set of decisions through which a board manages AI use by teachers and staff responsibly: which tools are in use, what has been decided for each tool and on what condition, whether a teacher sees that decision while typing, and whether the board can show that it works. The general picture is in our introduction to AI governance; at school, almost every decision comes down to whether pupil data can go in.
What do teachers use AI for?
Three kinds of work. Lesson materials: an assignment at three levels, a test with a mark scheme. Feedback: marking an essay, wording a report comment. Administration: making a letter to parents easier to read, writing up a parents' meeting.
Then there is a fourth category that nobody thinks of as AI use: the AI features inside learning materials the school already has, such as adaptive practice programmes. And a fifth: work finished at home on a personal account, because the school laptop was already shut. A policy that only describes the first three misses the fastest-growing part.
What pupil data goes in with the text?
For lesson materials, usually none. For feedback and administration, almost always some: a name, a class, a grade and, in the more serious cases, the reason for a concern. A support plan holds data about the health and behaviour of a minor, which is special category data under the GDPR. A report comment saying "has struggled to concentrate since the divorce" is part of a pupil's file too.
The task can nearly always be done without it. Softening a report comment works just as well with "the pupil" as with a name, and summarising a support plan is only responsible in an environment the school has an agreement with. Which data should always come out of a prompt is covered in what you can and cannot share with AI. So the board's decision is not "AI, yes or no", but per tool: is it allowed, and with which data?
What do the Privacyconvenant and the data processing agreement settle, and what do they leave open?
The Privacyconvenant is the agreement between the Dutch education sector and suppliers of digital learning materials on how pupil data is handled, and it comes with a model data processing agreement. The school is the controller, the supplier is the processor, and no school has to negotiate those terms from scratch.
That covers what the school procures, not a free chatbot a teacher uses on a personal account: no supplier has signed the Privacyconvenant, there is no data processing agreement and the school often does not know the tool is in use. With AI tools, personal and business accounts differ widely, and everything that belongs in a data processing agreement for an AI tool applies to a school in full. The Privacyconvenant answers "which suppliers do we have agreements with?", not "which tools are being used?".
What does the AI Act require of a school?
Two things. Since February 2025, Article 4 of the AI Act has required every organisation that uses AI systems to work on the AI literacy of the people using them. For a school, that means teachers know which tool they are using, which data does not belong in it and where to report a new tool. What counts as sufficient for each role is covered in our overview of AI literacy in 2026.
The second carries more weight but affects fewer schools. Annex III lists education as a high-risk area, including AI used for admission and for evaluating learning outcomes. A teacher who asks a chatbot to prepare a lesson is not using a high-risk system. A school that uses AI to grade essays or admit pupils does need to ask the question, together with the provider of that system, because the classification depends on the intended purpose. That distinction is worked through in our article on when an AI application counts as high risk; how it applies in practice varies by school and by system.
How do you divide this between board and school?
| Question | Board | School |
|---|---|---|
| Which AI tools are in use | One register for all schools, fed by what is actually opened | Reports a new tool, including one inside a learning material |
| Is this tool allowed, and with which data | The decision per tool: allowed, limited use, no decision yet, not allowed, with the condition on pupil data | Applies the decision in day-to-day practice |
| How does a teacher use AI in a lesson or a test | A framework, and room to choose | The teaching choice, per subject and per assignment |
| Do teachers know what has been decided | The notice when someone opens a tool | Discussion within the department |
| Does it work | Figures per board and per school, without names | A reason to adjust something |
What happens at a board with nine schools?
The board's policy in the opening example names two tools: Microsoft 365 Copilot is allowed, and ChatGPT is limited use without pupil data. When the board first looks at which AI services are being opened in the browser, a lesson materials tool turns up at two schools, a transcription tool used in parents' meetings at one, and at all nine a presentation tool that nobody had recognised as AI. None of the three is in the register, and the transcription tool records conversations in which concerns about a child are discussed.
That is not an incident: the teacher using the transcription tool saves an hour a week, and nobody had said it was not allowed because nobody knew the tool existed. A policy that names two tools is three short a year later. What follows: assess three tools, choose a status for each, check whether an environment with a data processing agreement can replace the transcription tool, and show that alternative as soon as someone opens the old one.
BeeSensible matches a catalogue of over 1,000 AI tools against the use observed in the organisation: only the domain, at most once a day per device, with no user ID and no content. In the Inbox, a board administrator sees which tools have no decision yet and records one; if the board has turned notices on, a teacher sees the decision in the browser as soon as they open the tool. BeeSensible does not block anything on its own initiative; whether a teacher can carry on to a tool that is not allowed after seeing the notice is a setting the board chooses. This is what the register looks like:
