How2Revise
Parent login Start revising

Home / How2Revise and AI

Current status

Free-text AI is not open.

The current public revision reset follows a fixed, human-written sequence. It does not send a pupil’s target to an AI, generate subject answers or offer an AI conversation.

Effective and last updated: 30 August 2026

No pupil prompt is being served by an AI feature.

Accounts, saved pupil profiles, payments, uploads and free-text AI remain disabled until named human owners, qualified UK reviews and documented safety tests are complete. A description of a planned feature is not a claim that it is available.

The only permitted future purpose

Teach the method. Preserve the pupil’s thinking.

If an AI feature passes its gates, it may help a pupil practise a revision method using a short piece of material they already have. It must not become a general chatbot, content library or answer engine.

What a pupil would always be told

“This is an AI revision tool. It can make mistakes. Check important information in your trusted source or with a teacher. Do not enter your name, address, school, phone number, password or private information.”

The proposed boundary

Useful revision support—not an unrestricted assistant.

May do after approval

  • Explain a reviewed How2Revise revision technique.
  • Break one revision task into a short, finite plan.
  • Ask for the pupil's own attempt before offering a small hint.
  • Turn a short, lawfully used extract into private practice prompts.
  • Tell the pupil to compare anything important with their original source or teacher.

Must not do

  • Writing homework, coursework, essays or answers to submit.
  • Acting as a friend, therapist, counsellor, teacher or emergency service.
  • Teaching unverified curriculum facts, predicting grades or diagnosing ability.
  • Live web search, pupil-to-pupil messaging, public sharing or image/audio uploads.
  • Advertising, paid steering, emotional dependency, secrecy, guilt or endless engagement.
  • Training a general AI model on pupil prompts, outputs or safeguarding concerns.

Privacy requirements before AI can open

  • A child-specific data-protection impact assessment and lawful-basis review must be approved.
  • The AI provider, subprocessors, processing locations, retention and international-transfer safeguards must be documented.
  • A provider must be contractually barred from using pupil input or output to train a general or commercial model.
  • Ordinary pupil prompts and responses must not be stored by How2Revise by default or copied into analytics, marketing or developer logs.
  • The model must not receive adult billing identity, a school, exact location or a long-term emotional/ability profile.
  • A pupil and parent must receive clear, age-appropriate information and a route to report, challenge or request human review.

Safety requirements before AI can open

A provider filter alone is not enough. Input and output moderation, personal-data minimisation, age-aware testing, safe fallbacks and a tested incident route are required. Clear danger, abuse, grooming or self-harm language must stop the revision flow and show a trusted-adult, Childline and 999 route.

A trained safeguarding lead and deputy must provide real cover. How2Revise must not claim continuous human monitoring unless it actually exists. Read the safeguarding statement.

Release test

Every model, system instruction and safety-filter version must be recorded and tested for ages 10–12, 13–15 and 16, including UK language variations, misspellings, accessibility needs, multi-turn attacks and harmless phrases that could be blocked by mistake.

One unresolved critical failure blocks release. Critical failures include harmful instructions, sexual content involving a child, continuation after a clear emergency disclosure, emotional dependency or secrecy, submission-ready assessed work, private-data leakage, unsafe parent notification, lost safeguarding alerts or training on pupil content.

An AI cannot approve itself.

The accountable director, privacy lead, trained safeguarding lead, UK solicitor, accessibility reviewer and independent AI-safety reviewer must approve their own areas. This page does not say those approvals have happened.

Official guidance we use

Using guidance does not mean a regulator, government department or awarding body has approved How2Revise.

Questions or reports

The branded privacy and safeguarding inboxes are not active yet. Do not send personal information to the planned addresses. Use the postal and official routes on Help and Safeguarding. For immediate danger in the UK call 999.