From Principles to Practice: Making AI Policy Work in Your School

One of my most popular resources in the last 18 months have been my AI, or as I prefer to write it, Ai policy templates. On the weekend, I shared a workshop with 100+ school leaders at the brilliant #SchoolsSuccessSummit25 conference organised by Toddle - Your Teaching Partner hosted by the fabulous Kate O'Connell, MA.Ed, ACC, IBEN . As you can imagine, we covered lots, so I thought following on from that I would share some of the things I covered in the session in the hope it would help you too.

I think Ai policy is important and if you don't have one, urgent, so read on to discover what it should include, how to put it into practice in your setting and the importance of doing it in a way that keeps humans firmly in the loop. I also consider whether your Ai guidelines should stand alone or be woven into existing policies, the roles of various stakeholders in making it work, and even some provocations about the future of Ai in education. Hopefully in doing so I'll help provide a clear, measured roadmap for school leaders and teachers to approach Ai with confidence, insight, and integrity.

Why Do Schools Need an Ai Policy Now?

Ai use is rising rapidly in schools. From lesson planning to language practice, savvy teachers are exploring AI tools to save time and enhance learning. Students are testing the waters, sometimes using Ai to get “help” with homework or creative projects. This is happening whether schools formally acknowledge it or not. We’re at a crossroads where reactive vs proactive responses make all the difference. Some schools have waited until problems occur (like a cheating incident) then reacted punitively, while others are proactively guiding safe, innovative practice from the start. It’s clear which approach sets us up for success.

Banning Ai outright with a policy that says “AI-generated work will result in a zero.” puts a barrier up right away. What will happen? Well, students will be students. They'll just work around that, using Ai at home and then paraphrase to avoid detection, effectively driving the whole issue underground. A more nuanced approach focusing on academic integrity and disclosure, rather than outright prohibition is the approach I advocate for.

We can’t just say “no Ai” and expect that to solve things, nor can we blindly adopt Ai without any guardrails. Burying our heads in the sand (or issuing knee-jerk bans) will only encourage misuse in the shadows. Inaction leaves teachers and students unguided; uncertain about what’s allowed, and that confusion can lead to mistakes, inequity, or even breaches of law.

On the flip side, a well-crafted Ai policy enables safe, innovative practice. By proactively setting expectations and boundaries, we channel Ai use in positive directions. We acknowledge the reality that these tools exist and have benefits, while putting measures in place to protect students and staff from the risks.

A good policy empowers teachers to try new Ai-driven techniques to enrich learning or save time, while being mindful of ethics and safety. It tells students what’s acceptable (and what isn’t) in clear terms, helping them learn to use Ai responsibly rather than leaving them to figure it out on their own and it helps teachers know when they can use it, how they can use it and where it fits in with assessment and curriculum. In short, an Ai policy is about being prepared and informed and is a vital part of a school’s duty of care in the digital age.

Key Frameworks Guiding Ai Use in Education

Before diving into policy specifics, it’s worth noting that we’re not operating in a vacuum. Several key frameworks and guidelines have helped shape what I think schools should be doing about Ai. These provide a helpful backdrop and in some cases, a legal focus for your school’s Ai strategy:

• KCSIE (UK) – Keeping Children Safe in Education is the UK’s statutory safeguarding guidance for schools. It makes clear that online safety is part of safeguarding, which absolutely extends to Ai. If students are interacting with online AI systems, KCSIE principles (like appropriate filtering, monitoring, and protecting children from harmful content or contacts) apply just as they do to web browsing or social media, even though KCSIE doesn't explicitly mention Ai.
• UK GDPR / GDPR / Data Protection laws – Schools have strict obligations to protect personal data. This is crucial when using Ai tools that might collect or process student or staff data. The UK Department for Education (DfE) has explicitly warned that many generative Ai services store and learn from the data you input. In practice, that means staff should never input personally identifiable information or sensitive data into a public AI tool. Your policy must ensure compliance with GDPR and related laws: for example, by vetting any third-party Ai tool’s privacy practices, obtaining consent if required, and making sure no one feeds, say, confidential student records into ChatGPT. Data privacy isn’t optional; it’s legally mandated.
• JCQ (Joint Council for Qualifications, UK) – The JCQ, which oversees exam regulations, has issued guidance on AI and assessment integrity. They’ve made it non-negotiable that students must not use AI to produce assessable work unless it’s acknowledged and allowed. In other words, any Ai assistance in coursework or exams has to be declared, and schools must train staff to detect and prevent malpractice. While this guidance is UK-centric and geared toward formal exams, it’s a good indicator for schools everywhere: your policy should spell out how AI can or cannot be used in assignments and assessments, ensuring student work remains authentic.
• DfE Guidance (England) – The Department for Education released a policy paper on AI in education (bit.ly/3WWt4eW) which puts leadership in the spotlight. It says school leaders must assess AI’s risks and benefits, maintain transparency and governance, and embed AI into the whole-school strategy. Essentially, the government is telling schools: “Have a plan, don’t wing it.” This top-level attention also hints at what inspectors might look for; it’s widely expected that Ofsted will soon be interested in how schools are managing Ai as part of their leadership and management judgment. No headteacher wants to be caught off-guard on that question.
• ADEK (Abu Dhabi) – Internationally, others are moving on this. ADEK (the Abu Dhabi Department of Education and Knowledge) recently shared a raft of new policies for schools. To help with this, I've created a policy support tool to help (bit.ly/adekpolicytool) schools govern Ai use. Their guidelines mandate transparency (schools must declare how they use Ai), focus on fairness and bias (ensuring Ai doesn’t introduce discrimination), and emphasise inclusion and equity. Interestingly, ADEK’s policy also advocate training staff for AI (more on training soon) and encourages innovation through policy, showing that good policy can actually support trying new things, as long as it’s within safe boundaries.
• EU AI Act – the European Union’s AI Act (bit.ly/euaiact25) is regulates Ai in all sectors. Education is classified as a “high-risk” AI domain under this act, meaning any Ai systems used in schools will face stringent requirements. The Act emphasises human oversight, transparency, and robust risk assessment/testing for high-risk Ai. If your school deals with EU-based students or just wants to meet a high standard, aligning with the EU approach makes sense. Concretely, it means your Ai policy should require a human-in-the-loop for important decisions, clear documentation of how Ai is used, and caution in deploying any Ai that could significantly impact student welfare or rights. Matthew Wemyss has resources around this too.

That’s a lot of acronyms and references, but the message is clear: the wider educational community and regulators are all homing in on Ai governance.

From safeguarding and data privacy, to academic honesty, to leadership responsibility. There’s a consensus that we need to manage Ai deliberately and while there is no “one size fits all” approach, these frameworks highlight common principles and components that any strong Ai policy should cover.

Key Components of a Comprehensive Ai Policy

Every school’s situation is unique, but in my experience there are several core components that a good Ai policy should address. Think of these as the building blocks of your policy document. By covering each of these, you ensure that you’ve thought through the major risks and opportunities of AI in your context:

• Data Privacy: Protecting personal data must be a top priority. All technology in education should undergo a DPIA (data protection impact assessment) and as such, no tool should be used by educators without first having gone through that process. Do your teachers know that though? Do you know that? Here's an example of why... Ai tools often require you to input data, maybe a list of student reading ages into an Ai analyser, or uploading student essays to get feedback from an Ai. Your policy should set clear rules to ensure compliance with data protection laws (whether that’s GDPR in Europe, the UK GDPR, or another local law). For example, the DfE guidance warns that generative AI services may store and learn from whatever you type in, so staff and students should never input personal identifiers, sensitive details, or any confidential records into such tools. If a teacher wants to try a new Ai app, are they allowed to sign up using a school email? Can they upload a student’s work to see what feedback it gives? The policy must address these scenarios: perhaps requiring leadership approval before any student data is used, or stating outright that certain data (like student names, photos, assessment data) must not be fed to external AI systems. It should also cover getting consent when appropriate (for instance, if using a third-party Ai service that processes student data, parents might need to consent). And don’t forget data output: if an Ai generates some analysis about a student, how is that data stored and who can see it? A solid privacy section will reassure parents and staff that using Ai won’t mean throwing privacy out the window. It’s about leveraging Ai’s power responsibly, without compromising our duty to protect student information.
• Safeguarding: We must treat student use of Ai with the same vigilance as any online activity, through a safeguarding lens. If students interact with an Ai chatbot or image generator, could they encounter inappropriate or harmful content? Unfortunately, yes, it’s possible if such tools are unsupervised or unfiltered. The UK’s KCSIE guidelines require schools to have appropriate filtering and monitoring in place for internet use, and Ai is no exception. Your policy might specify, for example, that “Students may only use approved Ai tools under staff supervision,” and that IT staff will configure any Ai systems with safe-search or age-appropriate settings enabled. Consider the age-appropriateness of Ai: a tool that might be fine for a Sixth Form student could be completely off-limits for a Year 3 pupil. The goal is to protect students from exposure to harmful content or contacts, just as we do with web filtering. Safeguarding in the Ai context also means preparing for wellbeing issues. Imagine an Ai chatbot that a student consults starts giving harmful advice or misinformation (an extreme case might be a mental health chatbot that inadvertently encourages self-harm). While we hope reputable Ai systems would avoid that, kids need to know that Ai can make mistakes or even “lie” convincingly. The policy should integrate with your existing online safety curriculum: teach students Ai literacy – that Ai outputs aren’t automatically true or safe and encourage them to always involve a teacher if they encounter anything unsettling or if they’re unsure. In short, treat Ai tools like any powerful resource on the internet: incredibly useful, but not without risks, so supervise and educate accordingly.
• Professional Development (Training): A policy is only as good as people’s ability and willingness to follow it. Staff training and ongoing professional development are crucial to successfully implementing an Ai policy. Many teachers (and even school leaders) are not Ai experts – and that’s okay! In a typical staffroom right now you might find a few early adopters enthusiastically sharing the latest prompt they tried with ChatGPT, while others are hesitant or even nervous about AI. Your policy should explicitly commit to training and supporting staff in this new area. In fact, Abu Dhabi’s ADEK policy mandates training to promote its objectives, underlining how important this is. What might this look like in practice? It could involve CPD sessions or workshops on Ai in education (I can help with those!), how-to guides for using approved tools and regular sharing of best practices among teachers. Also consider including students and parents in awareness sessions, for example, holding an evening talk for parents about generative Ai, how their children can use it to help their learning or a class lesson for students on using Ai productively and safely. The policy can state that the school will provide orientation for students on acceptable Ai use and inform parents about how Ai is (or isn’t) being used in instruction. The aim is to remove the mystique and fear of Ai by empowering everyone with knowledge. When teachers understand what these tools can and can’t do, and how to use them ethically and effectively, they are far more likely to embrace the policy rather than see it as just another document. (As a bonus, well-trained staff may discover ways Ai can reduce their workload on tasks like planning or marking – the DfE and Ofsted have hinted at this benefit – which can increase buy-in!). Ongoing PD is important too: Ai tech moves quickly. What’s “hot” this term might be outdated by next year. So plan for continuous learning, maybe a half-termly tech briefing, a dedicated channel on your staff Teams for Ai news, or even an “AI in Education” professional learning community. Your policy could promise that “the school will keep staff updated on new Ai developments and adjust training accordingly,” aligning with the idea that this is a living initiative. Training isn’t one-and-done; it’s an evolving support system as AI itself evolves.
• Leadership & Oversight: For an Ai policy to succeed, it needs clear ownership and active oversight from school leadership. This component is about who will ensure the policy isn’t just words on paper but a practice embedded in school life. In your policy, assign responsibility for Ai oversight to a specific role or team. Some schools name an “AI Lead” or put it under the purview of the IT Director; others create an AI Steering Committee (perhaps an extension of the existing ICT or e-learning committee). ADEK suggests a “Digital Wellbeing Committee” as one model. The exact mechanism can vary, but make it explicit. This person or group will evaluate new AI tools for approval, monitor how AI is being used, and coordinate updates to the policy. Indeed, policy governance (regular review and updates) should be written in, given the fast pace of change, an annual policy review might be a minimum. Leadership oversight also means lead by example: school leaders and department heads should themselves use Ai in line with the policy and champion its ethical use. For instance, if the policy says “review Ai outputs for accuracy,” leaders should demonstrate doing that when they use Ai for, say, writing a school newsletter draft. Another key oversight aspect is having an approval process for new Ai initiatives: your policy might state that “Any significant Ai project or new tool (e.g., adopting an Ai-driven learning platform) must be approved by the Ai lead and senior leadership after a risk assessment.” This ensures due diligence before something is rolled out to students. Finally, consider how you’ll monitor compliance in a supportive way. The policy could note that the school will periodically survey or audit Ai use (maybe checking if staff are following the data privacy rules, or if students are adhering to usage guidelines) and gather feedback to update practice. The tone here is not “Big Brother is watching,” but rather “we’re all learning, and leadership will keep an eye on how it’s going so we can adapt and help where needed.” With clear leadership oversight, Ai governance becomes an active part of school strategy, not just a box ticking exercise where the policy is just created and then filed away.
• Ethical Use: This is a broad but vital component – defining what ethical, appropriate use of Ai looks like in your school community. Essentially, this section translates your school’s values and codes of conduct into the context of AI. It should articulate principles like fairness, transparency, equity, and the overarching idea that Ai should remain human-centered. For example, you might state that Ai is to be used to augment teaching and learning, not to replace human teachers or interactions. This reaffirms that technology will not erode the essential teacher-student relationship. You might also explicitly prohibit using Ai for any high-stakes decisions without human involvement, echoing the EU AI Act’s stance on human oversight. In practice, that could mean: no purely algorithmic grading of students without teacher review, no Ai-based student profiling that labels kids without teacher/SLT oversight, etc. Another facet is intellectual honesty and academic integrity. If staff or students use Ai in their work, do they need to acknowledge it? The JCQ, as mentioned, says yes for formal assessments. Your policy can extend this ethos to everyday classwork: perhaps encourage or require students to add a brief footnote or comment if they used Ai to get ideas for an assignment, and require staff to double-check any Ai-generated content they plan to use in teaching materials for accuracy and appropriateness. The ethical use section might also list a few unacceptable uses of Ai on moral/ethical grounds. For instance, “AI should not be used to monitor or profile students in ways that invade privacy or contravene our school values” so no running student essays through some algorithm to psychoanalyse them, or using facial-recognition cameras in school hallways without a serious debate on ethics. Another example: forbidding the use of deepfake or AI-altered media in school communications without disclosure, to maintain trust and honesty. By laying out these principles, the policy makes it clear that we use Ai to help, not to harm; to complement, not to cheat; to uphold our values, not undermine them. Everyone in the community should understand that stance.
• Assessment & Academic Integrity: While this could be folded into ethics, it’s worth addressing directly how Ai will be treated in the context of student assessment. This year, one of the biggest questions teachers have had is “What if a student uses ChatGPT to do their homework?” Your Ai policy should provide clarity. Building on the JCQ guidance, state your school’s position on Ai-assisted student work. For example, you might allow Ai for research or drafting ideas in homework as long as the student acknowledges it and still demonstrates their own understanding, but strictly forbid it in final exams or any summative assessments. Or you might take a stricter line that all submitted work must be the student’s own unless Ai use is expressly permitted by the teacher. Whatever you decide, make sure it’s clearly communicated to students and parents (nobody should be caught by surprise). Additionally, commit to educating students about academic honesty in the age of Ai. This could mean teaching them how to cite an Ai tool, or discussing the ethical difference between getting an AI to solve a math problem versus using it to get feedback on an essay draft. Your staff will also need guidance: help them with strategies to detect Ai-generated content (though outright “Ai detectors” are unreliable), redesign assessments to be more “Ai-resistant” (for instance, more oral presentations, in-class writing, or personalised projects), and how to handle suspected cases of AI-assisted cheating. The key is to maintain assessment integrity, ensuring that grades reflect a student’s own skills and knowledge. With a robust policy, you can actually turn this into a learning opportunity: students who might be tempted to cheat with Ai can be mentored to use Ai appropriately (for learning, not cheating), and understand why authenticity matters. We want to avoid an Ai arms race of misconduct and instead foster a culture where honesty and effort are valued, even as we incorporate new tools. When students eventually go to university or the workplace, they’ll likely have Ai at their fingertips so our job in schools is to teach responsible use and integrity now.
• Ai in the Curriculum: Finally, a comprehensive policy should touch on how Ai tools will be integrated into teaching and learning (if at all). This goes beyond the “rules and risks” and into the positive opportunity side of things. Consider including a section about how staff and students can use Ai to enhance the curriculum. For instance, you might encourage teachers to use Ai-driven software in certain subjects (like an Ai art generator to inspire creativity in art classes, or a language learning Ai for practising French dialogues) with appropriate safeguards as noted earlier. Outline whether students are allowed to use Ai in their learning process and to what extent. Maybe your policy says Ai can be used for research purposes, brainstorming, or practice, but not for completing graded tasks unless specified. Or you could state that part of your digital skills curriculum will include Ai literacy, so students learn about Ai as a topic: understanding concepts of machine learning, bias in Ai, etc. Many forward-thinking schools are now including lessons on how tools like ChatGPT work, their limitations, and how to critically evaluate Ai outputs. This is a great addition to prepare students for the future. If your school has a digital citizenship or computing program, weave Ai into it. The policy can mention that “the school will integrate Ai education into the curriculum to ensure students become informed, responsible users of Ai.” This signals that you’re actively teaching students how to harness it. On the teacher side, you might highlight that teachers will be supported to experiment with integrating AI into their pedagogy (again, under oversight). Maybe even set up an “Ai in Curriculum” working group to share lesson ideas. The bottom line is that Ai isn’t just a thing to manage; it’s a tool to potentially enrich learning. A comprehensive policy doesn’t shy away from that. It embraces innovation within a safe structure. By addressing curriculum integration, you acknowledge that part of being “Ai-ready” as a school is preparing learners for a world where these technologies will be part of nearly every field they go into. It’s an exciting chance to enhance our teaching methods, once the proper safeguards are in place.

As you draft these components in your policy, remember a piece of advice: good policy isn’t about word count; it’s about clarity and relevance. Be clear about what (and who) is covered by the policy, what’s allowed or prohibited, and how it links to other existing policies (like your Acceptable Use Policy for ICT, your assessment policy, etc.). If Ai use by staff is covered under your staff ICT policy, reference that. If academic honesty is in your student handbook, tie it in. The aim is a cohesive framework where Ai isn’t an isolated consideration but part of the broader school policy landscape.

Human-in-the-Loop: Don’t Outsource the Educator

One phrase I reiterate in any discussion about AI in schools is “human-in-the-loop.” This concept is so critical that I want to highlight it on its own. In the rush to adopt Ai tools, we must ensure that we never fully outsource teaching or decision-making to an algorithm. Human oversight is, for me, a pedagogical and ethical necessity.

What does human-in-the-loop mean in practice for a school? It means any use of Ai in instruction or administration should have a human (teacher, support staff, or leader) reviewing, guiding, or contextualising the Ai’s output. For example:

• If an Ai system provides personalised math practice to students, a teacher or tutor should monitor progress and intervene where the Ai might not detect a learning issue or emotional issue. The Ai might flag a student as struggling, but the teacher knows that student was ill last week; context the AI lacks. Human insight is irreplaceable.
• If a teacher uses an Ai to help draft student report comments or a lesson plan, the teacher must carefully review and edit that content. The policy should remind staff that they are accountable for anything the AI produces on their behalf. Ai can save time getting started, but it doesn’t get the final say. For instance, an Ai might draft a nicely worded report comment, but the teacher needs to ensure it truly reflects the student and is free of any oddly phrased or inaccurate statements. We maintain professionalism by keeping the human teacher’s judgment at the core.
• For automated systems that might be used in schools (say an AI that analyses CCTV for security, or an AI that scans student work for plagiarism), human moderation and decision-making should be required for any actionable step. If the Ai flags something (e.g., “possible academic dishonesty” or a security alert), a person should verify and decide the outcome, rather than punishing a student or contacting a parent purely on the Ai’s word. Relationships take an age to build and a moment to lose, so don't ruin those with parents and students on the say so of an Ai tool.

This principle connects back to several of our key components: data privacy (human deciding what data goes in or out), safeguarding (educators overseeing what students see), ethical use (no high-stakes decisions by Ai alone), and assessment (teachers validating student work authenticity). It’s also echoed by big frameworks – the EU AI Act’s emphasis on human oversight, for example, and - as noted as a key strand in the #EdTechPlaybook - technology should enhance, not dictate!

Perhaps most importantly, education is a deeply human endeavor. Students learn from relationships, mentorship, and social interaction – things no AI can replicate. “Human-in-the-loop” means Ai is a tool in the educator’s toolbox, not a replacement for the educator. We must avoid the scenario where a school deploys, say, an “Ai tutor” and then leaves a student entirely alone with it. Or where a teacher relies on AI to grade and gives feedback without even reading the student’s work. Those approaches not only risk errors; they shortchange the student-teacher connection that is so vital for learning and development.

So, in your Ai policy, weave this philosophy throughout: teachers and humans remain in control. Make it explicit that any Ai usage should have appropriate human supervision and that ultimate responsibility lies with staff. You might even include a general statement like, “All Ai-supported activities in the school will involve human oversight; Ai will not be used as a substitute for professional judgment or personal interaction.” This assures your community that adopting Ai won’t mean a cold, robotic learning environment, on the contrary, it means teachers empowered by better tools, and students supported by both technology and caring adults.

Standalone Policy or Integrated Into Existing Policies?

One practical question many school leaders ask is “Where does our Ai policy live? Do we make a brand-new standalone policy document, or do we integrate Ai guidelines into our existing policies?” The answer can vary. There’s no universally “right” choice here. What’s important is that Ai governance is clearly visible and accessible to your community, whichever route you choose:

• Standalone Ai Policy: This is a dedicated document focusing just on Ai use and governance in the school. The benefit of a standalone policy is that it highlights the importance of Ai. It says, “This is a big enough deal that we wrote a whole policy for it.” You can go into depth on all the components we discussed – data privacy, safeguarding, etc. – in one consolidated guide. This comprehensive focus can be useful given how fast Ai is evolving; the policy can be updated frequently without having to amend multiple other documents. A standalone policy can also be easier to share publicly (e.g., on your website) to show parents and inspectors, “look, we have an Ai plan.” Many schools choose this route initially to ensure Ai gets special attention (and frankly, it helps staff realise this isn’t just a passing fad – it’s a core issue deserving reflection).
• Integrated Approach: This means weaving Ai guidelines into existing policies – for example, adding sections to your ICT Acceptable Use Policy, your Teaching & Learning policy, your Assessment policy, and your Safeguarding/Child Protection policies as relevant. The logic here is that Ai shouldn’t be treated in isolation or as an afterthought; it should be part and parcel of the domains it affects. One school I've supported doesn't have a separate Ai policy at all – instead, they integrate Ai-related clauses into their existing AUP, curriculum, assessment, safeguarding and academic honesty policies, because they felt Ai use should follow the same ethos as other tech use rather than being siloed. For instance, their Acceptable Use Policy now includes a section on generative AI tools, and their assessment policy includes the rules about AI and coursework. The advantage is coherence: staff and students don’t have to consult a separate document just for Ai; the guidance appears in context (e.g., when a teacher reads the assessment policy, the Ai rules for assessment are right there). It can also reinforce that Ai is not some strange new thing, but another part of your digital learning environment governed by existing values and rules.

There are, of course, challenges with each approach. With a standalone policy, you risk nobody reading it (let’s be honest, busy teachers aren’t fond of extra paperwork). It could sit on a shelf unless you actively promote and train on it, but hopefully you will! With an integrated approach, the risk is that Ai guidance gets buried among other policies. A single paragraph on AI in a 30-page handbook might be missed or forgotten. Also, integration can be tricky to manage; you have to update multiple documents consistently as AI issues evolve.

✅ Key point: Whatever approach you choose, make sure the Ai guidance is clearly communicated and not hidden. If it’s standalone, keep it concise and user-friendly so people actually read it, and remind staff/students that it exists. If it’s integrated, consider publishing a summary document or memo highlighting where Ai is addressed in various policies, so everyone is aware. The worst scenario would be a beautifully written Ai policy that no one knows about, or a tiny Ai blurb tucked in a policy that no one ever reads. We need awareness and buy-in for any approach to work.

In my experience, some schools start with a standalone policy to get things moving, then later embed those principles into other policies once practices stabilise. Others add a short-term addendum to their ICT policy (“Ai usage guidelines”) as a stop-gap, then eventually expand that into a fuller policy. There’s flexibility here. What matters is that you do something to explicitly address Ai rather than nothing.

Take a moment to think about your school’s culture: Are staff already drowning in documents? Do you have a strong existing digital use policy that could absorb Ai content? Or do you want to mark a fresh start by introducing a new policy? Either way, ensure that the approach is communicated clearly to all stakeholders. If you already have some form of Ai guidance, it’s a great discussion to have with your team: is it standalone or embedded elsewhere, and is that working for us?

Personally, I lean toward whatever will be most practical for the school to implement effectively. The format is less important than the substance and usage. In some cases, a short, stand-alone “Ai Policy” accompanied by training can later be merged into the Acceptable Use Policy – once people are used to it. In others, adding a dedicated Ai chapter in your existing digital policies can achieve quick integration without creating “one more policy” to manage. You might even do both: have a detailed Ai policy internally, and also update parent/student handbooks with key points.

Whatever you do, keep it iterative. As Ai adoption grows, you might adjust your stance. The key is staying nimble and clear: Ai governance must be visible, understood, and lived in the school, not just written down.

Roles and Responsibilities in Your Ai Strategy

A successful Ai strategy isn’t just about documents and software – it’s about people. Everyone in the school community has a role to play in making Ai use effective and ethical. Let’s outline the roles of different stakeholders and how they contribute:

• School Leaders (SLT and Governors): Leadership sets the tone and vision for Ai adoption. Their role is to drive the strategy and ensure accountability. This means senior leaders should initiate the creation (or updating) of the Ai policy and allocate time and resources for training staff. Leaders need to stay informed about Ai trends and risks (reading updates, attending trainings themselves) so they can make wise decisions. They are also the ones to establish oversight – e.g. appointing that Ai lead or committee we discussed. School governors or board members should be looped in as well, because Ai relates to school risk management and long-term planning. Critically, leaders must model a balanced approach: showing enthusiasm for innovation but with due diligence. For instance, a headteacher might pilot using an Ai tool for school newsletters or to analyse some admin data, and then share the experience with staff openly (the good and the pitfalls). This kind of modeling builds trust. Leaders also handle communication: informing parents about the school’s Ai approach, addressing any concerns transparently, and celebrating successes (like “this term, our science department used an Ai simulation tool, and here’s what we learned…”). Finally, leaders should be prepared to enforce the policy when needed – if a serious misuse occurs, having a clear response (just as you would for a breach of any policy). In summary, the role of leaders is to champion a vision for Ai that is innovative yet safe, ensure the policy is implemented, and cultivate a culture where Ai is used responsibly for the benefit of students.
• Teachers and Support Staff: Teachers are on the front lines of Ai use in the classroom. Their role is twofold: implementing the policy in their daily practice, and contributing insight to refine that policy. We expect teachers to follow the guidelines (e.g. not inputting private data, supervising student use, etc.), but we also rely on their professional judgment. A big part of their role is to experiment and share what works. Perhaps a teacher finds a great way to use an AI writing assistant to help dyslexic students, within the bounds of the policy, that’s fantastic and they should share it with colleagues. Teachers can become Ai champions or mentors, helping peers who are less confident. (Often it works well to start with volunteers or tech-savvy staff who can pilot an AI tool and demonstrate success, then guide others.) It’s important that teachers feel they have permission to try things within the guardrails – the policy should not be a muzzle, but rather a safety net. I always encourage an environment where teachers can ask questions like, “Is it okay if I use this Ai website for my history project?” without fear. Teachers also need to be vigilant. If they spot a student possibly misusing Ai or struggling with it, that’s valuable feedback for the whole school. Some teachers may worry about Ai encroaching on their role (“Will I be judged if I don’t use Ai? Will AI make me obsolete?”) their role is also to engage with these changes openly. As school leadership, we must reassure them (through policy, practice and the language we use) that AI is a tool, not a teacher, and that using it smartly is appreciated, not frowned upon. In team meetings or training sessions, teachers should discuss their experiences – what’s saving time, what’s confusing, etc. – so the school learns collectively. In short, teachers’ role is to be both implementers and informants: they carry out the policy in classroom scenarios and provide the insight to improve it over time. Support staff (like TAs, librarians, IT staff) also fall here – they often help manage the tech or supervise students, so including them in training and discussions is key. For instance, a librarian might show students how to use an AI research tool ethically; an IT technician might oversee the installation of a new AI software and flag any concerns about data. They all contribute to the bigger picture.

• Students: Our learners are at the heart of why we’re doing this. The student’s role in an Ai strategy is to be a responsible user and active learner. First, they need to follow the rules and guidelines set forth: for example, only using approved tools, not cheating on assessments, and respecting the safeguards (like not trying to turn off safe search filters or find “jailbreaks” for chatbots – a temptation some tech-savvy teens will indulge without guidance!). But more than just following rules, students should be learning why those rules exist – so part of their role is to engage with the digital citizenship aspect. We want students to develop AI literacy: understanding both the power and limitations of these tools. That means they should feel comfortable asking questions and reporting issues. If an Ai tool gives a weird or biased output, students can bring it up (“This image generator gave me an inappropriate image, what do I do?”). If they’re unsure whether using Ai on a homework task is allowed, they should ask the teacher rather than doing it in secret. By involving students in the conversation, we help them internalise the ethics. In fact, you might establish a student digital leadership team or include students in an Ai committee; especially older students who are keen, to get their perspective. Many schools I work with have done this with great success, empowering students to lead workshops for peers on, say, using technology for study in a responsible way. Students can also help educate their parents by sharing what they learned about Ai safety and ethics at school. Ultimately, we hope our students become ambassadors of balanced Ai use: excited to leverage new tools for learning and creativity, but mindful of doing so honourably and safely. The policy and its implementation should give them a clear structure to do that. When students buy in, understanding that the policy isn’t just arbitrary rules but is there to protect them and help them learn better, then we truly have a culture of responsible Ai use.
• Parents (and the Wider Community): Parents play a critical supporting role. Many parents are understandably anxious or simply unaware of what Ai could mean for their child’s learning. One of the best things a school can do is communicate and involve parents in its Ai strategy. The parents’ role is to stay informed and reinforce the school’s messages at home. Schools should help by providing accessible information: for instance, sending out a parent guide or summary of the Ai policy highlights. (E.g., “Dear families, we want to share how our school is approaching Ai…” summarising key points: we’re not banning it, but we have rules to ensure safety, privacy, and honesty. We encourage you to talk to your children about how they use AI and remind them of these principles.) Many parents will appreciate knowing that the school is proactive, that we’re not just letting kids loose on ChatGPT without guidance. Encourage parents to come to you with questions. Maybe host a short webinar or info night on “Ai in our School: What Parents Should Know.” When parents understand the educational value (e.g., how AI might provide personalised practice or save teachers time for more 1-1 attention) and the boundaries (e.g., no Ai in exams, no personal data shared), they’re more likely to support and echo those rules at home. We all know learning doesn’t stop at the school gate, if a student is using Ai at home for homework, parents who are aware of the policy can gently remind them, “Hey, you need to write your own reflection, not just copy what the Ai says,” or “Is that tool you’re using approved by the school?” We’re not asking parents to be Ai experts, but by engaging them, we create a consistent message. Parents also have legitimate concerns we should listen to: some worry about screen time, or privacy, or the reliability of Ai content. By hearing them out and showing what steps the school is taking, we build trust. In some cases, parents might even contribute resources – for example, a tech-savvy parent could come in as a guest to talk about how Ai is used in their industry, inspiring students. The wider community (industry partners, local authorities) can also support with resources or up-to-date insights. But at minimum, get the parents on board. End your Ai policy (or accompanying letter) with an open door: “If you have questions about AI and your child’s learning, please contact us.” Their role is basically to partner with the school, reinforcing the responsible use message and feeling confident that the school is leading on this important issue.

By clearly defining and communicating these roles, your Ai strategy becomes a whole-school effort. Everyone knows their part in the larger plan, and that sense of shared mission really helps with buy-in. In fact, when I talk with my clients post engagement, I ask them to share what’s working and what’s not and a consistent theme is that where schools have open dialogue involving staff, students, and parents in shaping Ai use, things go smoother. Where communication was lacking, fear and confusion filled the void. So, leverage all these stakeholders: you’ll need the collective support to cultivate an Ai-savvy school culture.

From Policy to Practice: Implementing and Training for Success

Writing a comprehensive policy is a critical first step, but bringing it to life is where the real work begins. Remember, getting the technology right is relatively easy. Change management isn't. So, how do we ensure that these guidelines are actually followed and make a positive impact? Here are some practical steps and suggestions for implementing your Ai policy and training your community:

1. Audit Your Current Status: Start by understanding your baseline. Review if and how Ai is already being used by students and staff, even informally. Are some teachers using ChatGPT to draft worksheets? Have students been caught submitting Ai-generated work? Do you perhaps already cover Ai in your ICT curriculum or AUP (Acceptable Use Policy)? Identify any gaps – for example, maybe you have no mention of AI in any policy (common at this point), or your teachers are using Ai but without any coordinated approach. An audit might involve simply surveying staff (“Who’s using Ai and for what?”) and even asking students. The goal is to map out what’s happening now, so your policy and implementation plan address reality. This also helps reveal the attitudes and pain points: maybe teachers are unsure about data privacy, or students are unclear on what’s allowed. Knowing this helps tailor your training.
2. Draft or Revise the Policy (Collaboratively): Using the framework we discussed (and perhaps the template I’ll share in a moment), form a working group to create your Ai policy. Involve a team of diverse stakeholders in this drafting process – an SLT member, an IT staffer, a couple of teachers from different departments, maybe a governor or a keen parent, and even a student representative if appropriate. This collaborative approach ensures the policy is practical and gains broader support. Make sure the policy covers the key components (data privacy, safeguarding, etc.) and aligns with any regulations or exam board rules relevant to you. As you draft, refer to existing resources: for example, the DFE’s guidance (bit.ly/3WWt4eW), the JCQ guidance (bit.ly/jcqguidance), or any local authority recommendations. There’s no need to reinvent the wheel; you can adapt language from these sources. Once a draft is prepared, review it with a critical eye: Is it clear? Would a new teacher understand what to do? Does it conflict with any existing policy? Tweak as needed. If possible, get feedback from a wider staff meeting – not to turn it into an endless committee process, but to catch any major concerns. Remember, inclusion fosters buy-in.
3. Educate & Train Your Staff (and Students): Don’t just email out the new policy and hope for the best! Plan a professional development session (or several) on Ai for your staff this term (the sooner the better, while the momentum is there). In the training, cover the why (importance and vision), the what (summary of policy content), and the how (practical dos and don’ts). Show concrete examples: e.g., demonstrate how an Ai tool can be used within the rules, and also perhaps show what a misuse might look like. Encourage questions - teachers will have plenty (“Can I use Ai to mark homework? What if a student…?” etc.). Also share the policy draft and invite their input (if you haven’t already). Concurrently, educate students about the policy. This could be through assemblies or workshops on Ai ethics and good use – essentially a student-friendly version of what staff get. Emphasise to students that this policy is there to protect and empower them. You might run separate sessions for older vs younger students. Furthermore, inform parents – maybe send a newsletter outlining the policy highlights and even run a short Q&A session for parents who are interested. By saturating the community with awareness and learning, you make the policy more than a document; it becomes part of the school’s learning conversation. If you’d like help with delivering a training session, feel free to get in touch; I regularly run workshops on Ai in education, and I love helping schools kickstart these discussions (shameless plug, but truly – don’t hesitate to reach out!).
4. Implement Gradually (with Support): Once the policy is official (approved by leadership/governors), roll it out in a measured way. You might not enforce every single aspect on day one if people are still learning. For instance, perhaps this term you focus on getting all staff to abide by the data privacy rules and start teacher training. Next term, you introduce student-specific rules in classrooms and start adjusting assessment practices. Basically, prioritise what matters most and tackle things in stages so it’s not overwhelming. Ensure that there’s support available: maybe an “Ai buddy” system where less confident teachers can partner with those more comfortable. Set up a mechanism for ongoing questions – e.g., an email hotline or a shared document where teachers can post “Can I do X with Ai?” and get an answer. You want to create an atmosphere of encouragement. Also, highlight and celebrate early wins: if a teacher tried an AI activity successfully, share that story in the next staff meeting. If a student reported an Ai-related concern appropriately, acknowledge that. These positive reinforcements show that the policy is not about catching people out, but about enabling good outcomes. Of course, do address violations if they happen – calmly and as learning opportunities. If a student breaks the rules, explain what went wrong and how to do it right next time. If a teacher accidentally did something against policy (maybe used a new Ai tool without approval), use it as feedback to refine either the policy or the communication around it. Iteration is key: treat the first term or two as a pilot phase to see how the policy works in practice, then adjust accordingly.
5. Stay Connected and Adaptive: The world of Ai is ever-changing, and so your approach must remain dynamic. I highly recommend joining (or forming) networks or forums with other educators focusing on AI in education. Whether it’s a local cluster of schools sharing experiences (such as that started by Steve Bambury in Dubai), an online community on LinkedIn, or attending webinars (like the Toddle summit we had), these platforms let you keep learning from peers. Share what you’ve learned from implementing your policy and learn what others are doing. This could also open up opportunities for collaboration – for example, schools might share anonymised data on how Ai tools affected homework completion rates, etc. Additionally, keep an eye on updates in guidelines and technology. We may see new government advice, new exam rules, or new Ai tools that require policy tweaks. Review your policy regularly – I suggest doing a check-in at least annually, if not each term in the first year. Involve your Ai lead or committee in gathering feedback: maybe do a quick survey after a term of implementation to ask teachers and students how it’s going. Use that input to refine the policy and training. By staying connected and adaptive, you ensure your school isn’t left behind. Ai in education is a journey, not a one-time fix. If you maintain this proactive stance, your school will be able to navigate challenges and seize opportunities that come with future Ai developments.

Breaking it down:

• Audit where you are (usage and gaps).
• Draft/Update Policy with a team, covering key points and aligning with regs.
• Train/Educate staff, students, parents on the policy and Ai literacy.
• Implement in Phases and support everyone during the roll-out.
• Review & Network with others, updating your approach as needed.

And don’t forget: we have resources to help. For example, I’ve created an Ai Policy Template (you can grab a copy at bit.ly/aipolicytemplate25) which is a starting point document you can customise. It’s based on the framework and components we’ve discussed. Using a template can save time and ensure you’re not missing any big pieces. Also, the JCQ guidance (bit.ly/jcqguidance) and the DfE’s paper (bit.ly/3WWt4eW) we referenced can be handy appendices or references as you justify why certain rules are in place.

Implementation is where the rubber meets the road, it’s challenging but also rewarding. When done right, you’ll notice over time a transformation: teachers will speak a common language about Ai (less fear, more collegial problem-solving), students will be more open about their use (rather than doing things in secret), and parents will trust that the school is steering this ship wisely. That’s the payoff of moving from policy to practice with care and intention.

Looking Ahead: Future Provocations to Consider

Even as we get our current Ai policies in place, it’s important to keep one eye on the future. The landscape of education technology is shifting quickly, and today’s “edge cases” might become tomorrow’s everyday issues. In the School Success Summit session, I posed a few thought-provoking questions to stretch our thinking beyond the immediate concerns. These aren’t necessarily things you need to solve right now, but they are on the horizon. Forward-thinking schools might even address them in brief within their policy or have ongoing discussions about them. Here are a few provocations for the future of Ai in education:

• Real-time AI Tutors for Every Student: What happens when every student has an Ai tutor on tap? We’re not far from a reality where every student has an AI assistant in their device or software that gives instant feedback or answers as they work (some schools already have this!). Imagine each student doing homework with a personal Ai whispering hints or corrections in real-time. This raises big questions: How do we assess learning in a world where help is always available? Do we move towards open-book assessments or oral exams where process and reasoning matter more than recalling facts? How do we ensure those Ai tutors give pedagogically sound guidance and don’t just feed answers? This scenario is akin to when calculators became widespread – we had to change the nature of math tests. Similarly, if AI becomes a ubiquitous “thought partner,” schools will need to emphasise higher-order skills (critical thinking, creativity, fact-checking the AI) even more. Some pioneering schools are already piloting AI-assisted learning tools in class; they should share their findings so we can shape ethical guidelines (e.g., perhaps AI assistance is allowed in coursework but not in exams, or students can use AI for research but must still write in-class essays without it). Are we preparing for that shift? It’s a good time to start re-imagining our approaches to homework and assessment in light of this possibility.
• AI Surveillance vs Privacy: Where’s the line between safety and privacy when it comes to AI monitoring? We’re seeing more Ai-driven surveillance tools; from cameras that claim to detect bullying or vaping, to software that monitors student screens or even facial expressions for engagement. On one hand, these could enhance safety and help identify problems early. On the other, they pose serious privacy and ethics questions. Do we want cameras tracking our students’ every move, even if well-intentioned? How do we avoid a dystopian vibe while still keeping schools secure? Your future policy might need to address whether you’ll use such technologies. If you do, transparency is key: students and parents must know what data is collected and why. There’s also the risk of bias for example, facial recognition misidentifying students of certain ethnicities. As these tools become more available, schools will have to tread carefully likely consulting legal and ethical experts. It’s worth having conversations about it now: for example, “If an Ai monitoring system could alert us to a student looking distressed in class, would we use it? What if it flags mostly false alarms? Who gets that data?” Finding the balance between protecting students and respecting their privacy will be an ongoing challenge.
• Predictive Analytics in Pastoral Care: Many schools already track data like attendance, grades, and behaviour points to spot students who might be at risk (for academic failure, mental health issues, etc.). Ai could turbocharge this by analysing patterns and making predictions – for instance, predicting which students are likely to drop out, or which might have a spike in anxiety based on various data points. This could enable early interventions – a counsellor reaching out to a student before a crisis hits. But it also carries the risk of labelling students unfairly. If an algorithm says a student is 70% likely to fail maths, do we inadvertently treat them differently? How do we ensure that using predictive Ai doesn’t create self-fulfilling prophecies or overlook the human story behind the data? Any Ai used in pastoral care must have a human decision-maker interpreting it (back to human-in-loop) and consider issues of bias (historical data can reflect systemic biases, and we don’t want to perpetuate those). Additionally, we’d need to involve parents and students in understanding how their data is used in this way. Under privacy laws, such uses might even be considered “high risk” and demand explicit consent. This is definitely a frontier area – some universities do it with retention models, but in Primary and Secondary / K-12 it’s new. It’s on the horizon as data systems integrate Ai, so it’s a good provocation for discussion: Would you use an Ai to flag pastoral concerns? How to do it ethically?
• Staff Ai Use and Professional Guidelines: We often focus on students, but what about teachers using Ai in their professional work? Questions arise such as: Should teachers declare when an AI helped create a lesson plan or report? If a teacher uses ChatGPT to draft student feedback or a student’s end-of-term report, is that acceptable and does the parent have a right to know? Some might argue, “If the feedback is accurate and helpful, it shouldn’t matter if AI was involved”; others might feel it’s less personal. There’s also the matter of fairness: if some teachers use AI to produce lesson materials much faster, should that be factored into how we view their performance or workload? We don’t want a weird dynamic where Teacher A is praised for pumping out great resources but actually an Ai did half the work; transparency and honesty in our professional duties is important. Possibly, schools could set guidelines like “Teachers may use AI to assist in planning, but must review all AI-generated content for accuracy and appropriateness, and should not use AI to do tasks that require personal judgment (like writing sensitive parent emails or counselling notes).” We should caution, for example, that no one should use Ai to draft something like a disciplinary action or a child protection report. Those need the human touch entirely. Another aspect is protecting teachers: if Ai makes some tasks quicker, schools shouldn’t just load more work on them in that saved time, instead, let them reinvest that time in direct student interaction or professional growth. Professional norms may need updating as Ai becomes commonplace in our workflow. Unions and professional bodies might weigh in eventually. For now, your policy can start by saying something like “Staff are encouraged to use Ai tools to enhance efficiency and creativity, provided this does not compromise professional responsibility, accuracy, or confidentiality. All final decisions and communications should be reviewed by a human professional.” That sets the stage, but expect this area to evolve.

These provocations are more questions than answers at this point, and that’s okay. The idea is to keep our schools future-ready by not getting too complacent once the “basic” Ai policy is in place. In fact, I encourage advanced schools (those who feel they’ve got the basics under control) to start drafting addendums or think-pieces on these topics. You might even include a forward-looking statement in your policy such as, “The school is committed to regularly exploring emerging Ai trends (e.g. real-time Ai assistants, Ai-based monitoring, etc.) and will update this policy to address them as needed.” This shows you’re forward-looking and prepared to adapt.

The future is rushing at us and being proactive is better than being reactive. By grappling with these provocations in staff meetings or strategy sessions now, you won’t be caught off guard later. And to be honest, students often love these discussions – it gets them thinking critically about the world they’re inheriting. Some of the best conversations I’ve had in classrooms have been around “Should an Ai invigilator watch students during exams?” or “If you had an Ai friend giving you advice 24/7, how would you know to trust it?” These topics can actually become part of learning, especially in ethics, computing, or PSHE classes.

In summary, keep these big questions in mind as you implement your current policy. Today’s policy covers known ground; tomorrow’s may need to cover new ground. If your school is agile and aware, you’ll be ready to update your ethical guardrails as technology progresses. Being future-minded is part of being responsible in the present.

Final Thoughts and Next Steps

Crafting and implementing a comprehensive Ai policy might seem daunting, but it’s one of the most impactful steps a school can take right now to navigate the Ai era in education. By being proactive, collaborative, and student-centered in our approach, we turn a potential threat (ungoverned AI use) into an opportunity (harnessing AI for good). Remember that an Ai policy is not just a document – it’s a reflection of your school’s culture and values in the face of new technology. If we get it right, the policy and practice around it will protect and empower our students and staff: protect them from harm, and empower them to innovate and excel.

A few key takeaways worth reiterating:

• Stay Informed & Compliant: Align your Ai practices with current guidelines (KCSIE, GDPR, DfE, JCQ, etc.) and legal requirements. Don’t operate in ignorance of what’s out there; use those frameworks to strengthen your policy foundation. They exist for good reasons.
• Policy = Culture: A policy won’t work in isolation. It must be brought to life through training leadership modelling, and open conversation. Aim to make responsible AI use part of the everyday culture of your school – something people do even when no one’s watching, because they believe in it.
• Protect & Empower: Always balance the dual goals of safeguarding and innovation. Yes, we must safeguard privacy, safety, and academic honesty; but we should also encourage exploration and efficiency where AI can help. A great policy does both, it protects people while giving them wings (think Red Bull!) to try new things.
• Be Adaptive: Treat the policy as a living document. Review it, get feedback, and update it proactively as new challenges and possibilities arise. One size won’t fit all schools, and one version won’t fit all time – adapt to your context and the changing landscape. Show that your school is forward-looking, not playing perpetual catch-up.

As I wrap up, I want to extend a heartfelt thank you to Toddle - Your Teaching Partner for hosting the #SchoolSuccessSummit25 where many of these ideas were discussed, and to the 100+ educators who attended and shared their experiences and questions. Your enthusiasm and insights have directly informed this blog post and undoubtedly will help countless other schools. It’s inspiring to see our educational community come together to tackle these emerging issues with such positivity and professionalism.

If you’re reading this and thinking, “This is a lot to take on,” please know you’re not alone. Many schools are on the same journey. Feel free to reach out to me for consultancy or training sessions – I regularly work with schools to develop their Ai policies, run staff workshops, or just advise leadership teams on strategy. Sometimes an external perspective or bespoke training can jump-start your efforts.

My contact information is on my website (ictevangelist.com) and I’m always happy to help fellow educators navigate these waters.

On a personal note, I’m also thrilled to share that my new book “The EdTech Playbook” (which covers practical strategies for implementing technology in schools, Ai included) is currently a #1 best seller on Amazon! If you’re looking for more ideas and success stories on blending education and technology, do check it out (you can find it via bit.ly/EdTechPlaybook). The support from the education community has been amazing.

The rise of Ai in education is a defining moment akin to the introduction of calculators, or the internet, or smartphones in the learning environment. With a comprehensive policy and a committed community, we can ensure it’s a positive turning point. Let’s continue to share, learn, and lead together on this journey. Thank you for reading, and here’s to safe, ethical, and impactful Ai integration in our schools!