In a trust, AI decisions made school by school produce inconsistency, duplicated spend and uneven risk. A steering group fixes that — here's how to set one up that actually functions.
Why a trust needs one
Without central coordination, each school negotiates its own tools, writes its own policy of varying quality, and carries its own unassessed risk. Meanwhile the trust misses the obvious advantages of scale: one procurement, one policy framework, shared training and shared learning. A steering group is the mechanism for capturing those.
Who should be on it
- A trust executive lead with authority to make decisions
- Your DPO — essential, not optional
- IT / network lead for what's technically feasible and safe
- A DSL representative for safeguarding implications
- Two or three headteachers across phases, so it isn't all central
- A classroom practitioner — the group loses touch fast without one
- A trustee or governor for oversight
Keep it under about eight. Larger groups discuss; smaller groups decide.
The most common failure: a steering group made entirely of central team and IT, with no classroom voice. It produces policies that are technically sound and practically unusable, and schools quietly ignore them. One practitioner in the room prevents most of that.
What it should own
- The trust-wide AI policy and staff acceptable use position
- Which tools are approved, and the process for requesting new ones
- Data protection position and DPIAs
- The training and CPD plan across schools
- Pupil-facing AI use and its safeguarding implications
- Monitoring what's actually happening in schools
- Reporting to trustees
A realistic first six months
Meeting 1 — where are we?
Audit actual current use across schools. Most trusts are surprised — informal use is always further ahead than the centre thinks.
Meeting 2 — data and risk
DPO-led. Establish the data position and what needs a DPIA before approving anything.
Meeting 3 — policy
Agree the trust framework and how much local flexibility schools get.
Meeting 4 — tools
Decide the approved list and whether to procure centrally.
Meeting 5 — training
Plan CPD across schools — see what CPD staff need.
Meeting 6 — review and report
What's working, what isn't, and report to trustees.
Central framework, local flexibility
The balance that works is a firm trust-wide position on the non-negotiables — data, safeguarding, approved tools — with schools free to decide how AI is applied to their own workload priorities. Mandating identical practice across a trust rarely survives contact with individual schools' realities.
Keep it moving
Half-termly is about right. Less often and it drifts; more often and busy people stop attending. And give it a genuine decision-making remit — a group that only advises will find its recommendations quietly unimplemented. See AI governor briefings.
Setting up AI across a trust?
We help MATs build a framework that works centrally and locally. AskColin gives schools low-cost monthly support with one-to-one help whenever you need it — so when you're stuck on a real job at 8am, there's someone to ask. Practical, jargon-free, built around your team.
Get one-to-one AI support for your schoolFrequently asked questions
Who should be on a MAT AI steering group?
A trust executive lead with decision-making authority, the DPO, an IT lead, a DSL representative, two or three headteachers across phases, a classroom practitioner, and a trustee for oversight. Keep it under about eight — larger groups discuss, smaller groups decide.
What should a MAT AI steering group own?
The trust-wide AI policy and acceptable use position, approved tools and the request process, data protection and DPIAs, the training plan across schools, pupil-facing AI use and safeguarding, monitoring actual practice, and reporting to trustees.
What makes an AI steering group fail?
Being made entirely of central team and IT with no classroom voice — it produces technically sound, practically unusable policies that schools quietly ignore. Also meeting too rarely so it drifts, or having only an advisory remit so recommendations go unimplemented.
Should a trust mandate identical AI practice across schools?
No. A firm trust-wide position on non-negotiables — data, safeguarding, approved tools — with local flexibility on how AI is applied to each school's workload priorities works better. Mandating identical practice rarely survives contact with individual schools' realities.