A policy tells people what to do. A governance framework tells you how you'll know it's actually working. For AI in primary schools, that doesn't need to be complicated — it needs four moving parts that connect to each other.
The four parts of an AI governance framework
- Policy — the rules (see our AI policy guide).
- Training — staff can only follow a policy they understand; a one-off INSET session isn't enough on its own.
- Risk management — a living risk register reviewed at the same cadence as your other safeguarding documents.
- Governance oversight — a named governor or committee that receives a short AI update at least once a term.
Who should own this?
In most primary schools, day-to-day ownership sits with a deputy head or the DSL, with the headteacher accountable overall and governors providing scrutiny — the same pattern most schools already use for safeguarding and GDPR. You don't need a new committee structure; you need AI added as a standing item to an existing one.
What "good" looks like after one term
A school with a working governance framework can answer four questions without hesitation: what's our policy, who's trained, what's on our risk register, and when did governors last discuss it. If any answer is "I'm not sure," that's the part of the framework to build next.
Keeping it proportionate
The goal isn't to slow staff down. A framework that's too heavy gets ignored. A framework that's too light doesn't protect pupils or the school. Aim for something a governor could explain in two minutes and a teacher could follow without a manual.
Key takeaways
- Four parts: policy, training, risk register, governor oversight.
- Fold AI into an existing committee rather than creating a new one.
- Test your framework by asking: could a governor explain it in two minutes?