What is AI safeguarding in a school context?

It is the set of arrangements that keep children safe when AI tools are used by or around them: approved tools with age limits, rules about what pupil information may be entered, filtering and monitoring that covers AI services, human oversight of any decision about a child, and a clear route from an AI-related concern to the designated safeguarding lead.

Who is accountable for AI safeguarding in a school?

Safeguarding accountability does not move because the technology changed. The designated safeguarding lead remains responsible for concerns about children, the data protection lead for pupil data, and the head and governors for the policy that permits a tool in the first place. AI use should appear in the existing safeguarding and data protection reporting, not in a separate IT process.

Should children use AI chatbots at all?

Used within age limits, with an approved tool and adult oversight, AI can genuinely help with revision, explanation and accessibility. The risks come from unapproved consumer tools with no filtering, from emotional reliance on companion apps, and from pupil data going somewhere nobody agreed to. Decide the tool list deliberately rather than letting it happen by default.

My child told a chatbot they were struggling — what should I do?

Treat it as a disclosure. Talk to your child calmly and without blame, keep any screenshots, and contact the school's designated safeguarding lead. If you believe your child is at immediate risk, call 999. Childline on 0800 1111 is free and confidential for the child themselves at any hour.

Can an AI tool be used for wellbeing check-ins?

Only as a prompt to reach a person, never as the response. Any wellbeing feature used with children should say plainly that it is not a counsellor, signpost a named adult and national helplines, and let staff see patterns that need a human conversation. Our own wellbeing check-ins are built that way and remain free.

Does the school need a DPIA before using AI with pupils?

Almost always, yes. Processing children's personal data with a new technology is exactly the higher-risk case a data protection impact assessment exists for, and it should be completed before the pilot, not after. Record the outcome, the mitigations and the review date.

Back to home

For schools, colleges and parents

AI safeguarding for schools and parents

Children are already using AI tools, usually before any policy catches up. This page gives you a short risk checklist you can download, plain escalation guidance for when a child may be at risk, and the questions parents should be able to ask and have answered.

Download the checklist (CSV)Escalation guidance

The eight risks that come up most often

Work through these with your designated safeguarding lead. The downloadable checklist expands them into twelve checks with owners, evidence and a review date.

Common AI safeguarding risks, what each looks like in practice, and the action to take.
Risk	What it looks like	Action
R1Pupil data entered into an unapproved tool	Reports, SEND notes or behaviour logs pasted into a consumer chatbot to 'save time'.	Approve a short tool list, state what may never be entered, and treat any breach as a data incident on the day it is found.
R2A disclosure made to a chatbot instead of a person	A child tells an AI companion about harm at home, and nobody hears it.	Teach pupils that AI tools cannot help in an emergency, and make sure every tool used in school signposts a named adult.
R3Harmful or age-inappropriate output	Self-harm, eating-disorder or sexual content surfacing in an unfiltered general tool.	Ensure AI services are covered by filtering and monitoring, and check age limits on every tool before pupils use it.
R4Over-attachment to an AI companion	A pupil prefers the chatbot to friends or staff, or uses it late at night.	Watch for withdrawal, set usage boundaries at home and school, and route concerns to the designated safeguarding lead.
R5AI-generated bullying or fake images	Edited or generated images of pupils or staff circulating in group chats.	Handle as bullying or, where images are sexual, as a child protection matter — escalate immediately and preserve evidence.
R6Decisions about a child made on AI output alone	Predicted grades, exclusions or referrals drafted by a tool and signed without review.	Require a named human decision-maker for every decision that affects a pupil.
R7Academic malpractice	Assessed work produced by a tool with no pupil understanding behind it.	Teach and publish what acceptable AI use looks like per subject, before the coursework deadline rather than after.
R8Parents unaware of what is in use	A tool becomes routine in lessons with no communication home.	Tell parents which tools are used, why, and how to raise a concern.

Escalation guidance

If a child is at immediate risk of harm, call 999 now. Nothing on this page replaces your local safeguarding procedures — it is designed to sit alongside them.

Immediate risk of harm

If a child is in immediate danger, contact the police on 999 and the designated safeguarding lead. Do not wait to gather AI logs first — record what you saw afterwards.

A disclosure or a safeguarding concern

Report to the designated safeguarding lead the same day, through the school's normal safeguarding system, not through IT. Note the tool involved, what was said and any screenshots, and follow local authority referral thresholds.

Sexual images of a child, generated or edited

Treat as a child protection matter. Do not forward or copy the image. Report to the designated safeguarding lead, who will involve the police and, where appropriate, the Internet Watch Foundation or Report Remove.

Pupil data in the wrong place

Report to the data protection lead the same working day, record it in the incident log, contain it (revoke access, delete where possible), and assess whether it is reportable to the ICO within 72 hours.

Harmful output from an approved tool

Record the prompt and response, remove the tool from pupil access until reviewed, report to the provider, and log it so the approved tool list is updated at the next review.

Parents: where to start

Talk to the child first without blame, keep the evidence, then contact the school's designated safeguarding lead. For urgent help outside school hours, Childline (0800 1111) and the NSPCC helpline (0808 800 5000) are available, and 999 for immediate danger.

Childline: 0800 1111 (free, confidential, for children). NSPCC helpline for adults worried about a child: 0808 800 5000. Emergency: 999.

Six questions parents can ask the school

Which AI tools does my child use in school, and what age are they approved for?
What pupil information is allowed into them, and who decided that?
Are these tools covered by the school's filtering and monitoring?
Who is the designated safeguarding lead, and do they know AI tools are in use?
How would the school know if a tool gave my child harmful advice?
What does acceptable AI use look like for homework in each subject?

If you are choosing a study tool for your own child, our SAMU GCSE companion is built for oversight: parents and educators can see activity, wellbeing prompts signpost a human, and the free accessibility and wellbeing tools never sit behind a paywall.

AI safeguarding questions from schools and parents

What is AI safeguarding in a school context?: It is the set of arrangements that keep children safe when AI tools are used by or around them: approved tools with age limits, rules about what pupil information may be entered, filtering and monitoring that covers AI services, human oversight of any decision about a child, and a clear route from an AI-related concern to the designated safeguarding lead.
Who is accountable for AI safeguarding in a school?: Safeguarding accountability does not move because the technology changed. The designated safeguarding lead remains responsible for concerns about children, the data protection lead for pupil data, and the head and governors for the policy that permits a tool in the first place. AI use should appear in the existing safeguarding and data protection reporting, not in a separate IT process.
Should children use AI chatbots at all?: Used within age limits, with an approved tool and adult oversight, AI can genuinely help with revision, explanation and accessibility. The risks come from unapproved consumer tools with no filtering, from emotional reliance on companion apps, and from pupil data going somewhere nobody agreed to. Decide the tool list deliberately rather than letting it happen by default.
My child told a chatbot they were struggling — what should I do?: Treat it as a disclosure. Talk to your child calmly and without blame, keep any screenshots, and contact the school's designated safeguarding lead. If you believe your child is at immediate risk, call 999. Childline on 0800 1111 is free and confidential for the child themselves at any hour.
Can an AI tool be used for wellbeing check-ins?: Only as a prompt to reach a person, never as the response. Any wellbeing feature used with children should say plainly that it is not a counsellor, signpost a named adult and national helplines, and let staff see patterns that need a human conversation. Our own wellbeing check-ins are built that way and remain free.
Does the school need a DPIA before using AI with pupils?: Almost always, yes. Processing children's personal data with a new technology is exactly the higher-risk case a data protection impact assessment exists for, and it should be completed before the pilot, not after. Record the outcome, the mitigations and the review date.

Written by Andy Wake, Founder and Managing Director of Automated Governance Systems Ltd. Practical guidance for use alongside your statutory safeguarding duties — not legal advice.