AI / Acceptable Use of AI Policy
The three things a school AI policy has to balance
Nearly every argument about AI in a school is one of three considerations beating the other two. Naming them makes the argument shorter, and settling them in a policy means it does not have to be had again every time somebody suggests a new tool.
Last reviewed 16 October 2026
Benefit
There is a real one, and pretending otherwise costs credibility with staff. Administrative load is a genuine driver of people leaving teaching, and a share of it — drafting, rephrasing, summarising, the forms — is reducible.
The test worth applying is whether the benefit is to the pupil or only to the adult’s workload. Both are legitimate; they justify different amounts of risk. Saving a teacher two hours a week is worth a modest exposure. It is not worth processing children’s medical data through a service nobody has assessed.
Privacy
The school remains the controller for pupil data whatever tool a member of staff chooses. That does not move because the tool was free or because the decision was made at a kitchen table on a Sunday.
Privacy usually loses these arguments quietly rather than loudly — not by being overruled but by nobody asking where the data goes until after the tool is in use across three year groups.
Fairness
The one most often left out. It shows up in two places: pupils who have these tools at home and pupils who do not, and models that perform worse for some pupils than others.
Anything that sorts, scores or flags pupils needs this question asked explicitly, because the cost of getting it wrong falls on the pupils with the least margin — which is the opposite of what a school is for.
Settling it in the policy
The point of writing these down is that the next tool does not restart the argument. A policy that names the three and says how they are weighed gives the person evaluating a new platform a test to apply: what is the benefit and to whom, what data does it touch, and who could it disadvantage.
That belongs in an adopted document with a review date, not in the memory of whoever happened to be in the room.
Giraft
The policy that settles the argument once.
Giraft takes your AI policy through drafting, consultation and Board approval, tracks staff acknowledgement against the version in force, and brings it back for review while it is still accurate.
Book a Giraft demo