Using AI to Write School Reports: What DfE Guidance Means for Pupil Data
A practical guide to using AI for school report writing in England without putting pupil data at risk: what the DfE guidance says, a safer workflow, and questions to ask any AI tool.
Save hours on UK report comments
Generate National Curriculum-aligned comments in seconds. Pre-filled with a working example so you can see results immediately.
Try It Now →Pre-filled example • 1 token • 30 seconds
By the School Reports UK team – built for teachers in England
Plenty of teachers in England now use AI to help draft school reports. The Department for Education is broadly supportive of AI reducing teacher workload, but it is clear about the conditions: protect pupil data, and keep the teacher in charge of every judgement. Here is what that means for report writing in practice.
What the DfE says
The DfE’s policy paper on generative AI in education, and the support materials it published for schools in 2025, make three points that matter most for reports:
- Personal and sensitive data should not be entered into generative AI tools unless there is a lawful, secure basis for doing so. Report writing is exactly where that risk arises.
- Human oversight is essential. AI can produce inaccurate or biased text, so teachers must check the output.
- AI does not replace professional judgement. The assessment of a pupil is the teacher’s.
All of this sits on top of existing law. Report comments are personal data under the UK GDPR and the Data Protection Act 2018, and your school remains responsible for how that data is handled. Always follow your school’s own AI and data protection policies.
The general chatbot problem
Pasting a class list with names, SEND information or behaviour notes into a general AI chatbot is the scenario the guidance warns against. Even without names, a detailed description of a pupil’s circumstances can make them identifiable.
A safer workflow
- Never use names. Work from a pupil code or initials that only you can decode, and add the real name in your school’s MIS.
- Keep sensitive information out entirely. No diagnoses, safeguarding notes, pupil premium status or family circumstances. They do not belong in a report anyway.
- Give evidence, not identity. “Can explain equivalent fractions using a bar model” is what the AI needs. Who the child is, it does not.
- Check every draft. Is the achievement level right? Is the next step genuinely right for this pupil? Would the parent recognise their child?
- Make it yours. Edit the wording so it sounds like you. Parents notice when every report in the class reads the same.
How School Reports UK is designed around this
- You identify pupils with a code or nickname, never a real name. Comments come back with a name placeholder you replace yourself.
- Only the de-identified details you type are sent for generation. See our privacy policy for exactly what is processed and by whom.
- Prompts are built around how schools in England report: the expected standard, EYFS judgements, GCSE grades and British English, and they avoid codes and invented data.
- You review and approve every word. Nothing is sent to parents by us.
Questions to ask about any AI report tool
- Do I need to enter pupil names or other identifying details?
- Which AI providers process the text, and is it used to train their models?
- Can I delete my data, and how long is it kept?
- Does it write in British English using the language schools in England use?
- Has my school approved it under our data protection policy?
Sources
Ready to write your report comments?
Join School Reports UK to generate personalised, National Curriculum-aligned comments in minutes instead of hours.
Generate Example Comment →Pre-filled example • 1 token • 30 seconds
UK Report Comments, Written in Seconds
Generate in 30 seconds • British English • Reception to Year 13
School Reports UK turns your notes into parent-friendly report comments in the language schools in England use, ready to paste into SIMS, Arbor or Bromcom.
Choose a token packJoin UK teachers saving 10+ hours per reporting cycle.