AI Clarity Audit
The evidence KCSIE 2026 expects a school to hold on AI, written for governors.
Since 1 September, Keeping Children Safe in Education treats generative AI as a safeguarding risk, puts AI-generated images inside the definition of harmful content, and expects a documented review every year of whether your filtering and monitoring actually work. Most schools have the policies. Very few have the evidence. Three weeks, and you are holding it: every tool genuinely in use, including the one a department bought without telling IT, checked against what your filtering lets through, scored against six governance dimensions, and written up so a governor can read it.
from £2,400
Single school. Trusts priced per school, and it comes down. Annual re-review £950.
Right for you if: A head, DSL or trust lead who has to show governors what the school is doing about AI this year, and would rather show evidence than a policy.
Start with the auditNo pitch on the first call.
What changed on 1 September
Three duties, and all three land on the governing body.
- 01
Generative AI is now a named risk factor in the online-safety framework. Your online-safety and child-on-child policies have to say what that means in your school, not just that it exists.
- 02
AI-generated images, deepfakes and so-called deep nudes are now inside the definition of harmful content. Your DSL and every member of staff need to recognise them and know where they go.
- 03
Filtering and monitoring must be reviewed at least once every academic year — led by a member of SLT with the DSL and IT, checked on every internet-connected device in every location, and recorded. Governors are expected to see that record, and the review has to consider whether AI tools get through.
For academy trusts, the 2026 Academy Trust Handbook adds a fourth: the board's risk register has to cover the whole operation, not just the money. AI use by staff and pupils belongs on it. A policy template produces none of this. Evidence does.
How it runs
Step 01
Organisation profile
Twenty minutes from your exec lead. Who holds DPO, DSL, IT and CPD, who leads on filtering and monitoring, what your strategic priorities are this year, and what prompted the call.
Step 02
Anonymous staff readiness survey
Your whole staff body, genuinely anonymous. Results stay sealed until at least five people have replied, and the platform holds no route from an aggregate back to a person. Your staff can be told that plainly, which is why the free-text answers are usually the most useful page in the report — and why this is where the AI tool map comes from.
Step 03
Leadership and governance assessment
Fifty-three statements across leadership, adoption, capability, governance, safeguarding and infrastructure. Behaviourally anchored, so you are picking the sentence that describes your school rather than a number between one and five.
Step 04
Evidence and filtering review
Your data protection policy, acceptable use, online-safety and safeguarding policies and record of processing, read against what your staff say is actually happening and against what KCSIE 2026 now asks for. Then a live check of what your filtering lets through, AI tools included, on the devices pupils actually use — written up in the form the annual review record needs.
Step 05
Findings, roadmap and the governors' paper
A scored position on each dimension, the gaps ranked by risk, a sequenced plan, and a two-page paper for the board. Delivered live to SLT, then in writing.
What you keep
Things that outlast the engagement.
Everything below lives in your Client Hub and stays there after I have gone. You own it, and you can export all of it.
Scored position across six governance dimensions
The anonymous staff picture — what is in use, and the gap between leadership's view and the staffroom's
An AI tool map: every tool in use, who uses it, and what pupil data touches it
A supplier check on each AI tool, against the questions the ICO's 2026 edtech audit says schools should be asking
The annual filtering and monitoring review, recorded the way KCSIE 2026 expects
Policy gaps against KCSIE 2026 — online safety, child-on-child, acceptable use, staff conduct
Risk-register entries and a two-page governors' paper
A sequenced roadmap with owners and dates
What this is not
The audit ends at the map. Writing the policy and training four hundred staff is the programme below — though plenty of schools take the map and walk it perfectly well on their own. One thing does not end: the filtering review is annual by statute and the tool map goes stale in a term, so the re-review twelve months on is £950, and I will ask you to book it before I leave.
Or consider