AI Policy Library for UK Schools

ABOUT THIS PAGE

Using these AI guidelines

Pupils are already using AI for schoolwork. In research for Safer Internet Day 2026, 54% of UK 8 to 17-year-olds said they use AI to help with homework or studying, and half had seen people their age use AI to do their schoolwork for them. Only 31% of parents and carers thought their child had used AI for homework. The question for schools and trusts is no longer whether pupils will use it, but what the rules should be, who they apply to, and how to keep them current as the tools change.

This page pulls together what we've learned working with thousands of schools on those questions, written for schools and trusts in England, with pointers for Wales, Scotland and Northern Ireland. It covers whether you need a formal policy, what the DfE, KCSIE, JCQ and Ofsted expect, what a policy should cover, and real examples you can borrow from. Working in the US? See our US AI policy guide.

OUR TEAM'S SUGGESTIONS

Tips for forming your AI policy

Does my school or trust need an AI policy?

No law in England requires a stand-alone AI policy. But several things schools must already follow now expect your policies to cover AI:

Wales, Scotland and Northern Ireland each have their own guidance: the Welsh Government's guidance on generative AI in education on Hwb, the Scottish Government's guidelines and guardrails for AI in schools, and the Department of Education's circular on generative AI in Northern Ireland.

Even where it isn't required, teachers are asking for one. In an NEU survey of 9,408 teachers in English state schools, 76% said they now use AI in their day-to-day work, but 49% said their school has no AI policy for staff or pupils. In a Teacher Tapp survey for the Royal Society, only 15% of teachers in England said they had clear whole-school guidance on AI.

A policy doesn't have to be long or rigid. The DfE's support materials for leaders leave it to you whether to write a stand-alone AI policy or fold AI into the policies you already have. One approach is to keep the approved policy short (values, who's responsible, what's off-limits) and put the details that change often, like approved tools and classroom examples, in separate guidance that's easier to update.

What existing policies can we lean on?

Start with your child protection and online safety policy, behaviour policy, malpractice and non-examination assessment (NEA) policies, homework policy, data protection policy and privacy notice, and acceptable use agreements. Much of what an AI policy needs to say already belongs in them. The DfE's leadership materials say child protection, online safety and behaviour policies should be updated to reflect the fast-changing risks from AI.

Existing policies were usually written to prevent misuse, so they rarely say what pupils are allowed to do with AI. Your new rules should make clear whether pupils can use AI to revise, research a project, or get feedback on a draft, and when they need to say that they did.

WHAT TO INCLUDE

What to put in your AI policy

This checklist draws on the DfE's guidance, KCSIE and the JCQ rules. Use it when you draft or review your policy.

Set rules by key stage

The DfE leaves it to each school to decide how AI is used. Some schools use it only with staff, or only for admin. Others let pupils use it, but only in certain subjects, year groups or key stages. A rule that makes sense in the sixth form rarely fits Year 3, so set out what's allowed at each stage.

Check each tool's terms, too. The DfE's data protection guidance says some AI tools have age restrictions, and schools must check them before letting pupils use a tool.

Protect pupil data and vet your suppliers

Your policy should say that only approved tools may be used with pupils' personal data, and who approves them. The DfE recommends keeping personal data out of AI tools altogether unless your data protection officer has confirmed a tool is safe to use.

The DfE's generative AI product safety standards set out what it expects from AI products used in schools, from filtering and monitoring to data protection and child development. They were updated in January with new standards on cognitive development, emotional and social development, mental health and manipulation. Before approving an AI tool, ask the supplier:

Where your pupil data is stored and where the AI runs are separate questions: some tools store data in the UK or EU but send every prompt to a model running in the United States. Our comparison of AI platforms for UK schools goes through these questions for seven platforms.

Name the harms your policy covers

Cheating gets most of the attention, but many incidents are about safeguarding. KCSIE says its guidance on nudes and semi-nudes covers images that are digitally altered or wholly generated by AI, sometimes called "deepfakes" or "deep nudes". Taking or sharing nude images of anyone under 18 is a criminal offence, including AI-generated images. In the Safer Internet Day research, 60% of young people said they worry about someone using AI to make inappropriate pictures of them. Say plainly that creating or sharing fake images, audio or video of real pupils or staff is prohibited, and connect it to your behaviour and child protection policies.

Chatbots that act as a friend are a newer concern. In the same research, 41% of teenagers said people their age rely heavily on AI for emotional support. The DfE's product safety standards expect AI products not to imply they have emotions or an identity of their own, and to alert the DSL to patterns that suggest emotional dependence. Your policy should say what happens when a pupil brings a personal or mental health issue to an AI tool, and who is told.

Handle academic integrity without relying on detectors

For qualifications, the JCQ's AI guidance is the rulebook. Pupils must acknowledge any AI use in work submitted for assessment, and teachers must investigate when they doubt that work is a pupil's own. The JCQ lists AI detection tools but warns that their accuracy varies, including with a pupil's English language skills. It says they should be only one part of the evidence, and that teachers, who know their pupils best, are best placed to judge.

What works better is being clear, assignment by assignment, about how much AI help is allowed, and asking pupils to show how they used it. Many schools use a scale like Leon Furze's AI Assessment Scale so every assignment carries a clear level, from no AI to full AI use. Ofqual's AI in coursework resources include a lesson plan for talking to pupils about AI and coursework, and a briefing pack for senior leaders.

How to create your AI policy and guidelines

Every school's process looks a little different. Some of the schools that use Flint wrote a policy first, and others chose a tool and then wrote rules around it. These are the steps we see most often.

1. Sync with your school leaders. Bring together senior leaders and governors or trustees to agree on goals. The DfE's leadership materials suggest a link governor for AI or digital technology. Start by auditing which AI tools and built-in AI features pupils and staff can already reach, and whether you're updating existing policies, writing a new one, or both.

2. Gather an AI working group. Include teachers across subjects and key stages, people who are sceptical of AI as well as enthusiasts, your DSL, your data protection officer and your exams officer. The DfE's leadership materials also recommend listening to pupils, through surveys, interviews or pupil digital leaders.

3. Create a set of working guidelines. A small group can draft, using the checklist above. Then have teachers test the guidelines in real lessons and report what's unclear or missing.

4. Roll out and monitor reactions. Share the policy with staff, pupils, and parents and carers in plain language. Keep a way for people to ask questions and flag problems, whether that's department meetings, a named contact or a simple form.

5. Revise as necessary. Put a review date in the policy itself and keep it. The JCQ requires exam centres to review their malpractice policy every year, and new tools, new features in the tools you already use, and new DfE guidance will all require updates.

Further reading

• Generative artificial intelligence (AI) in education, the DfE's policy paper

• Generative AI: product safety standards from the DfE

• Using AI in education settings: support materials, free training from the DfE, Chiltern Learning Trust and the Chartered College of Teaching

• AI Use in Assessments from the JCQ

• How Ofsted looks at AI during inspection and regulation from Ofsted

• Generative AI and data protection in schools from the DfE