Responsible AI use in Pakistani classrooms: a one-page policy any school can adopt

Ask a staffroom who uses AI and you'll usually get a few raised hands and a few embarrassed ones. Teachers are already generating lesson plans, rewording explanations, and drafting parent messages with AI tools, mostly without any school-level guidance either way. That gap is where problems start: one teacher pastes a student's name into a chatbot, another circulates a worksheet nobody checked, a third is quietly told off for using a tool the principal secretly uses too. A short, shared policy fixes most of this, and it doesn't need a committee or a legal team to draft.

Good policies borrow from good frameworks. UNESCO's AI competency framework for teachers is built around five areas: a human-centred mindset, the ethics of AI, AI foundations and applications, AI pedagogy, and AI for professional learning, with three progression levels called Acquire, Deepen, and Create. You don't need to adopt the framework wholesale, but it gives a useful spine: a school policy should say something about human judgment, ethics, understanding the tool, classroom use, and staff learning. UNESCO's wider guidance on generative AI in schools adds two practical anchors: protect data privacy, and set an age limit of 13 for using AI tools in the classroom.

Here is a one-page structure, clause by clause. Clause one is purpose: say in a sentence why the school allows AI at all, for example to reduce repetitive preparation work so teachers spend more time teaching. A stated purpose makes every later decision easier, because you can ask whether a use serves it. Clause two is who may use what: staff may use AI for planning and preparation; students don't use teacher-facing generators; anything student-facing needs the principal's approval first. Keep it simple enough that a new joiner can read it in two minutes.

Clause three is the red line on data. Nothing that identifies a child goes into a general AI tool: no names, admission numbers, marks, photographs, health or family details. Everything a teacher realistically needs to ask, such as the class, subject, chapter, and learning outcome, is safe. This clause does more protective work than any other, so write it in plain words and repeat it at every staff briefing. Our companion post on AI data privacy lists the vendor questions that sit behind this rule.

Clause four is human review. Every AI-generated plan, worksheet, or quiz is a draft, and a named teacher is responsible for it before it reaches a classroom. That single sentence carries UNESCO's human-centred principle into daily practice: the tool suggests, the teacher decides. In the UNESCO Recommendation on the Ethics of AI, adopted by 193 member states in 2021, this is framed as AI systems not displacing ultimate human responsibility and accountability. A school policy can say the same thing in ten words.

Clause five is disclosure. Decide how the school labels AI-assisted material, for instance a small footer such as "drafted with AI, reviewed by the teacher," and when parents or students should be told. Clause six is what to do when something goes wrong: a wrong fact reaches a class, a tool leaks something, a colleague breaks the data rule. Name one person to report to, promise a no-blame first conversation, and say what happens next. Clause seven is a review date, because tools and rules both move quickly; six months is a realistic rhythm.

Some schools stop at the policy. The schools that make it work pair it with a short practice session: staff generate one plan with a real tool, review it against the board's actual learning outcomes, find one thing to fix, and discuss what they'd change. That exercise teaches the review habit better than any page of rules, and it starts moving teachers from Acquire toward Deepen in UNESCO's terms without anyone needing to name the framework.

Where does a product like Muallim fit into a policy like this? By design, we try to make the policy easy to follow rather than easy to bypass. Muallim is built by DIGIT Pakistan for teachers and school admins, not students. The Lesson Planner, the Worksheet Builder, and Quiz & Assessment open every result in an editor before anything is exported, so the human-review clause has a natural place to happen, and generation only needs a class, subject, and topic, which supports the data clause. That doesn't remove the need for your policy. It just means a policy and the tool aren't pulling in opposite directions. We're also a beta, so treat any tool, ours included, as something your policy governs and not the other way round.

If you want a fast start this week: copy the seven clauses above into a single page, put your school's name and a date at the top, adjust the wording to sound like your staffroom, and read it out at the next meeting. Ask each teacher to name one AI use they'll keep and one they'll stop. An imperfect page that everyone has heard beats a perfect one that lives in a drawer, and you can tighten it at the six-month review.