An AI Framework for Schools and Teachers: A Practical Guide to Getting It Right

1. Where schools actually are

Most schools are living in the gap between two facts. The first is that generative AI arrived in classrooms through the back door, on student phones, before any policy was written. The second is that teachers were handed the problem with almost no training. Surveys across 2024 and 2025 told the same story in country after country: a majority of secondary students had already used a chatbot for schoolwork, while only a minority of teachers said their school had given them any clear guidance on what to do about it.

That gap produces two bad reflexes. One is the ban, which fails because it is unenforceable and because it teaches nothing. Students keep using the tools at home, they just stop telling anyone. The other reflex is the open door, where the technology is waved through with no structure, and the predictable result is a rise in copied work and a quiet erosion of the thinking that school is supposed to build. Neither reflex is a plan.

The good news sits underneath the anxiety. The same research that shows students racing ahead also shows what teachers gain when they use these tools with intent. Time spent drafting lesson plans, differentiating a worksheet for three reading levels, writing a first pass of report comments, or building a quiz falls sharply. Teachers describe getting evenings back. The framework below is built to capture that gain while closing the door on the failure modes. It rests on a single principle: the point of school is the thinking, so AI belongs anywhere it protects the thinking and nowhere it replaces it.

2. The framework: five pillars

A framework is only useful if a busy vice-principal can hold it in their head. This one has five pillars. They are not sequential, they hold up the same roof at the same time, but they give you five distinct places to look when something is going wrong.

Literacy. Everyone in the building, staff first, then students, understands what these tools are and are not. That a chatbot predicts likely text rather than knowing facts. That it invents citations. That it reflects the data it was trained on, biases included. Literacy is the pillar that turns AI from magic into a machine with known limits, and a machine with known limits can be supervised.

Practice. The concrete, approved ways AI shows up in lessons. A teacher using it to plan and differentiate. A student using it to get feedback on a draft, to generate practice questions, or to argue against their own essay. Practice is where the value is created, and it needs examples, not slogans.

Integrity. The rules and, more importantly, the assessment design that keep AI from replacing the student's own thinking. This is less about catching cheats and more about setting tasks where outsourcing the work defeats the point.

Privacy. The line that protects student data. No child's name, image, grades, or personal circumstances go into a public AI tool. This pillar is often the one that carries legal weight, and it is the easiest to get badly wrong by accident.

Equity. The narrow pillar that holds the others honest. If AI help is only available to students with a phone, a data plan, and a quiet room at home, the school has just widened the gap it exists to close. Equity means the school provides supervised access so that capability does not track family income.

3. A staged rollout you can start this term

Schools that succeed do not switch AI on across the whole building in September. They move in stages, each one small enough to manage and honest enough to learn from. Here is the sequence I recommend to the schools I work with.

1. Form a small working group. Four to six people: a senior leader, two or three teachers who are curious rather than expert, someone who handles data or IT, and if possible a parent voice. This group owns the framework. It does not need to be the most tech-confident people in the school, it needs the most trusted.
2. Write a one-page position first. Before any tool is chosen, agree the school's stance in plain language a parent can read. What AI is for here, what it is not for, and who to ask. One page. If it runs longer, no one will read it, and a policy no one reads is not a policy.
3. Train the staff before the students. Teachers cannot supervise what they have never tried. Run two or three short, hands-on sessions where staff use the tools on their own real tasks, planning a lesson, marking a sample, building a quiz. Confidence comes from use, not from a briefing.
4. Pilot in a few classrooms. Pick two or three willing teachers across different subjects. Give them approved uses, a way to record what works, and permission to fail. A term-long pilot tells you more than a year of committee meetings.
5. Redesign the assessments that break. Use the pilot to find which tasks a chatbot can now complete in seconds, and change them. This is the real integrity work, and it is far more durable than any detector.
6. Publish, widen, and review. Share what the pilot learned, extend approved practice to more classes, and put a date in the calendar to review the whole framework. AI tools change every few months. A framework that cannot be revised is already out of date.

4. What teachers actually do: classroom moves

Frameworks live or die in the lesson. These are the moves I have watched work in real classrooms, phrased so you can try them on Monday.

Plan and differentiate. Ask the tool to turn one lesson objective into three versions of a task at different reading levels, then edit. You keep the professional judgement and lose the retyping. Teachers report this is where the largest time saving lives.

Make the AI the student. Have the class critique a chatbot's answer to a question. Where is it wrong? What did it leave out? What sounds confident but is unsupported? This flips the tool from an answer machine into an object of study, and it builds exactly the scepticism students need.

Feedback on process, not product. Let students use AI for feedback on a draft, then submit the draft, the AI feedback, and their revision together. You are now assessing how they responded to feedback, which a chatbot cannot do for them.

Socratic practice. Configure the tool to ask questions rather than give answers, so a student stuck on a maths problem gets a nudge, not a solution. Several school-safe tools now do this by default.

Teach the citation habit. Every time AI is used, students note where and how, the way they would cite a source. This normalises honesty and makes the tool visible rather than hidden.

The interactive tool below shows the shift teachers describe most often: where the hours go before and after AI enters planning and marking. Toggle between the two to see the reallocation, and notice that the total does not vanish, it moves toward the work only a teacher can do.

5. Guardrails: integrity, privacy, equity

Integrity without surveillance. Start by accepting that AI detectors do not work reliably. They miss edited AI text and, worse, they falsely accuse real students, and the false accusations fall hardest on students who write in a second language. A school that disciplines a child on a detector's say-so will eventually be wrong about a real person. The defence that holds is task design: more writing done in class, oral defences of submitted work, drafts and notes handed in alongside the final piece, and assignments that ask students to improve or challenge AI output rather than produce text from a blank page. Design the task so that cheating and learning are the same amount of effort, and most students will choose to learn.

Privacy as a bright line. The rule is short and absolute: no student's personal data goes into a public AI tool. Not their name attached to a grade, not their photo, not a description of a child's home situation for a report. If staff want AI help with sensitive work, they anonymise first or use a school-approved tool with a proper data agreement. Put this rule on the one-page policy in bold, because this is the pillar that carries real legal and safeguarding weight.

Equity as provision. If the school allows AI but only some students can reach it, the school has chosen to widen the gap. The answer is supervised access on school devices and school time, so that a student without a phone or home internet is not locked out of a skill their classmates are building. Equity is not a statement of values, it is a budget line and a timetable decision.

6. Measure your school's readiness

You cannot manage what you refuse to look at. The tool below is a live version of the five-pillar model. Set your honest level on each pillar and watch the readiness shape fill out. There is no passing score. The value is in the conversation the shape starts in your working group, and in seeing which pillar is holding the others back.

7. The mistake to avoid, and the one to make

The mistake to avoid is treating AI as an IT project. It is not about the software. The schools that struggle bought a tool and expected it to teach the framework for them. The schools that succeed treated it as a change in professional practice, supported their teachers through it, and accepted that the first term would be untidy.

The mistake worth making is starting before you feel ready. No school will feel ready. The tools move faster than any policy cycle, and a working group that waits for certainty will wait forever while its students carry on using AI with no guidance at all. Start small, start honest, and revise as you learn. A pilot in three classrooms this term teaches you more than another year of deliberation.

"The schools that get this right are not the ones with the biggest budgets. They are the ones that trained their teachers first and trusted them to lead." - Adrian Dunkley, AI Boss

8. The frameworks this builds on

This is a synthesis, not an invention. Schools do not have to start from a blank page, because a strong body of national and international guidance already exists. The five pillars above map onto the same themes those frameworks keep returning to: clear policy, AI literacy, data privacy, and phased integration. If you want to go deeper, these are the primary sources worth reading.

• AILit Framework, a joint European Commission and OECD initiative supported by Code.org, defines what learners should know and be able to do in the age of AI. ailiteracyframework.org
• TeachAI: AI Guidance for Schools Toolkit, ready-to-adapt policy language organised as iterative improvement cycles. teachai.org
• Australian Framework for Generative AI in Schools, six principles covering teaching, wellbeing, transparency, fairness, accountability, and privacy. education.gov.au
• The SEE Framework from AI for Education, a practical route to building generative AI literacy. aiforeducation.io
• Independent Schools NSW phased implementation framework, and the NCEE Framework for AI-Powered Learning Environments for system leaders.
• UNESCO guidance on generative AI in education, which recommends a minimum age of 13 for independent use.