School AI Policy Guide for UK Leaders and MATs

By Tomas Lerpiniere · Updated 2026-10-28

Headteachers, SLT and MAT leads need a short, workable AI policy — not a 40-page aspirational document. This guide outlines Red/Amber/Green learner rules, teacher marking expectations, DPO checks aligned with UK GDPR and DfE generative AI advice, and a fair protocol when staff suspect unacknowledged AI in assessed work.

In this guide

Feedback Flows tools in this guide

AI writing detector Writing fingerprints Reporting hub Students and classes

Red / Amber / Green learner uses

Band Homework / revision Coursework / NEA / EPQ
Green Explain a concept; quiz generation; idea lists with learner rewrite Only if awarding rules and centre policy allow, with clear acknowledgement
Amber Outline help that must be rewritten in the learner’s voice Limited planning support only; never paste AI prose into assessed drafts
Red Submitting AI text as original work when the task bans it Unacknowledged AI generation; fabricating sources; bypassing authentication duties

Publish examples in student language. Link teachers to JCQ AI guidance for NEA and coursework so subject teams interpret authentication stages consistently.

Teacher AI marking and feedback rules

  • AI may draft comments or criteria-aligned notes; humans release marks.
  • Prefer platforms with teacher review, audit trails and no training on learner work — see how we handle learner work.
  • Integrity tools must show quotes and allow recorded outcomes, not automatic fails.

Broader adoption habits sit in AI in schools: best practices and safe and responsible AI in classrooms.

DPO, ICO and data checks

Before rolling out detectors or marking AI, ask your Data Protection Officer: lawful basis, retention, sub-processors, international transfers, and whether learner content trains any model. Use the procurement checklist in questions to ask about student data and AI. Feedback Flows Ltd is ICO registered (ZC132812); free-tool pastes are short-lived, private and not used for training.

Department protocol for suspected AI misuse

  1. Teacher triage with writer context and sentence highlights via the free AI writing detector or full product.
  2. Optional second signals: authorship, citations, draft pack.
  3. Pastoral / subject lead meeting with quotes on screen — script ideas in how teachers spot ChatGPT in student essays.
  4. Record outcome; escalate only under JCQ/centre policy after authentication rules are checked.

Never write “failed by AI detector” into behaviour systems. Prefer evidence over scores and authorship workflows.

FAQ

Define Red/Amber/Green uses for learners, rules for teacher AI marking and feedback, data protection and DPO contacts, acknowledgement expectations, and a short protocol for suspected AI misuse that never relies on a detector score alone.

Green: permitted idea generation or explanation with disclosure where required. Amber: limited drafting help that must be rewritten and acknowledged. Red: submitting AI text as original assessed work, or using AI when a task bans it (common for NEA/coursework).

The school or college remains the data controller. Choose processors that do not train models on your learners’ work, document the lawful basis with your DPO, and keep teachers in control of final marks and integrity outcomes.

No — ban automated sanctions based on detectors. Allow assistive checks with quoted evidence, human review and recorded outcomes. Link staff to fair tools such as a free AI writing detector for initial triage.

At least each academic year, and after major JCQ or DfE updates. Collect teacher feedback each half-term on false positives, workload and learner understanding of the rules.

Use Feedback Flows articles on JCQ coursework AI rules, evidence over detection scores, writing fingerprints, and questions to ask about student data and AI, plus your DPO and local authority or MAT template packs.

Continue exploring

Feedback Flows tools for this workflow

Integrity AI writing detector A hybrid check with writer context for EAL and younger writers, highlighted passages you can quote exactly, and a place to record what you d...
Integrity Writing fingerprints Build a style profile from previous authentic samples, then compare a new submission. See how vocabulary, structure and habits align — evide...
Students and insight Reporting hub Intervention risk and mark risk views, breakdowns by year group and subject, AI cohort summaries, intervention logging and exports — built f...
Students and insight Students and classes Organise learners into classes, assign work, and keep marking and profiles in one place.