91����

Like it or not, AI has crept into school systems as well as daily life, appearing across search engines, management information systems, classroom tools and communication platforms. But left unchecked, its increasing use by pupils and staff can introduce safeguarding, ethical and data risks.

The potential benefits of AI, such as reductions in staff workload, improved accessibility and enhanced teaching and learning, can only be realised when AI use is structured and key challenges are addressed (such as inappropriate content, data misuse, misinformation and unsafe interactions).

Schools, therefore, need a clear, dedicated policy that sets out how AI, particularly generative AI, should be used.

But what exactly should your school AI policy cover?

Here are seven points you should consider as you put together a policy for your school:

Drawing up a school AI policy

1. Policy scope

Your AI policy should apply to all staff, pupils, governors, trustees and external users, and should cover teaching, learning, assessment, administration and communication, including off-site activity linked to the school.

It should also recognise that AI is embedded in many digital tools (such as search engines and email clients), not just standalone AI systems.

Your policy should define the purpose of use, when and how use is permitted, who is authorised to use AI, and any age or role restrictions.

It is likely that AI is already being used, so engagement with staff and pupils is essential to ensure a consistent and structured approach to adoption.

The intent should also be clear: safeguarding pupils and staff, ensuring ethical and responsible use, supporting learning and workload efficiency, and preparing pupils for an AI-enabled world.

2. Alignment with existing policies

Align your AI policy with your policies on safeguarding, data protection, behaviour, assessment, staff conduct and complaints.

Data protection is particularly important when it comes to transparency and automated processing.

Where AI is used in automated decision-making affecting pupils or families, schools are required to be transparent with parents and carers, ensuring that decisions do not reinforce bias.

Personal or sensitive data, such as names, dates of birth or information on special educational needs and disabilities, should not be entered into generative AI systems without consent and completion of a Data Protection Impact Assessment (DPIA), as .

There is a legal requirement to undertake a DPIA if you are using AI for anything likely to represent a “high risk” to individuals’ rights and freedoms - examples of processing that may constitute this are .

3. Roles and responsibilities

AI in schools is first and foremost a leadership issue. Clear accountability is essential.

Governors and trustees provide oversight. Senior leaders set strategy and approve tools, while the designated safeguarding lead manages safeguarding risks. The data protection officer ensures compliance and oversees DPIAs, and technical staff manage filtering and security.

All staff are responsible for using approved AI tools, protecting personal data, checking outputs for accuracy and bias, modelling safe practice and reporting concerns.

The policy should stipulate that human judgement remains central to decision making, and all AI outputs should be checked because they can be inaccurate or biased.

4. Approval of AI tools

Your school’s AI policy should set out that only approved GenAI tools are to be used, and new tools are not to be introduced without assessment.

The Department for Education’s outline the features that generative AI products should have to be considered safe for use in education. Leaders should use these standards to support evaluation of whether any tool is safe for use.

The standards, for example, require tools to have their own filtering to prevent harmful content, have an alert system for when a child accesses or generates anything concerning, and not to have manipulative design features.

To ensure that tools remain safe and compliant, review regularly and make approved lists accessible to all users.

5. Filtering and monitoring

Filtering and monitoring remain key safeguarding controls. It’s advisable to block generative AI by default, with only approved platforms and tools allowed, and to regularly test filtering across systems. You can access advice on how to perform filtering checks .

This is not a one-off task because risks evolve quickly. You can use data from your monitoring system to look at any concerning behaviour from AI use, such as children engaging in sexualised conversations with AI chatbots, using concerning language or seeking advice for mental health.

Monitoring this will lead you to questions around whether particular platforms are appropriate for children to access and whether your filtering is effective, and even thinking about how the curriculum needs to adapt to reflect how children are using generative AI.

6. Reporting and incidents

Clear reporting routes are essential. Consider your reporting processes for safeguarding concerns involving AI, misuse of tools and data protection incidents, as well as how these are recorded and escalated.

A good reporting process will have clear thresholds with explicit examples of what should be reported, how to report it and to whom.

7. Teaching, learning and staff development

Clear expectations on AI use help everyone. For pupils, this includes when AI use is permitted, how it’s to be acknowledged, expectations for independent work, and limits on overreliance.

For staff, this includes appropriate use in planning and resource creation, assessment boundaries and responsibility for checking outputs.

Ongoing professional development is essential, including AI tool training, bias awareness, data protection, evaluation of outputs and effective classroom application. Training works best when it is continuous, rather than one-off sessions.

AI is already shaping education, and ignoring it may increase risk rather than reduce it.

A clear AI policy provides consistency, safeguarding assurance and confidence for staff and pupils.

Alex Dave is safeguarding lead at charity LGfL. You can access LGfL’s AI Policy Toolkit, which supports schools in developing a safeguarding-led approach to AI,

��