AI transparency in schools: what to tell staff, parents, and students, and what to ask vendors

Parents rarely object to a teacher using a tool. They object to finding out afterwards that nobody mentioned it. The same is true in the staffroom, where a colleague who learns from a student that the school "uses AI now" feels sidelined rather than informed. Transparency is the cheapest part of responsible AI use and the part most often skipped, usually because nobody is sure what exactly there is to disclose. This post gives a concrete answer for each group.

Start with why it matters at the standards level. The UNESCO Recommendation on the Ethics of Artificial Intelligence, adopted by its 193 member states in November 2021, states that the ethical deployment of AI systems depends on their transparency and explainability. The OECD AI Principles, adopted in 2019 and updated in 2024, also list transparency and explainability among five values-based principles. You don't need to cite either document to a parent. But a school that can say "we tell you how we use AI, and here's where" is aligned with the direction both point in.

Begin inside the school. Staff should know which AI tools are approved, what may be entered into them, and who to ask when unsure. A simple register works: a table with the tool's name, what it's used for, who approved it, what data it receives, and the date it was last reviewed. It takes an hour to build and answers most questions before they're asked. It also protects teachers, because "the school approved this on that date for that purpose" is a much better position than "I thought it was fine."

For parents, disclosure should be short, specific, and non-defensive. A paragraph in the term newsletter is enough: teachers at our school use AI tools to help prepare lesson plans and worksheets; a teacher reviews everything before it reaches your child's class; we don't put student names or marks into these tools; here's who to contact with questions. Notice what that says and what it doesn't. It doesn't promise AI is perfect or claim it replaces teachers. It describes the actual practice and the safeguards, which is what a concerned parent wants to know.

For students, transparency has a second job, which is modelling good behaviour. If teachers label AI-assisted material, for instance with a small footer such as "drafted with AI, reviewed by your teacher," students see what honest disclosure looks like. That makes it much easier to ask them to declare their own use, as described in our post on preventing AI misuse. Children notice hypocrisy quickly, and a rule that applies only to them tends not to last.

Then there is transparency from vendors to the school, which is where a principal has the most leverage. Ask the vendor to say in writing which AI provider processes your content, whether your content is used to improve models, what personal data it collects, and how you can delete it. If the answers are vague, treat that as information. Google's Gemini API terms show why the specifics matter: they distinguish between paid and unpaid services in whether submitted content may be used to improve Google's products. A vendor built on such a service should be able to tell you which applies to you.

Explainability is the harder half. Most AI systems can't fully explain why they produced a particular sentence, and a school shouldn't pretend otherwise. What you can explain is the process around the tool: what went in, who reviewed the output, and what was changed. That's why "human in the loop" is not a slogan but a transparency practice. When a parent asks how a worksheet was made, "a teacher generated a draft, checked it against the syllabus, and edited three questions" is a complete and honest answer.

Here is how Muallim tries to meet its own standard. We're built by DIGIT Pakistan for teachers and school admins, and our Privacy Policy names Google's Gemini API as the AI system that generates lesson plans, worksheets, and quizzes, and says the details you enter are sent to Google for processing. Our public roadmap says plainly which features, such as an exam-paper generator or parent messaging, are not built yet. The Lesson Planner, the Worksheet Builder, and Quiz & Assessment open results in an editor before export, so the "who reviewed it" part of your answer to a parent is built into the workflow. We're still a beta, and we'd rather tell you that than let a polished page imply otherwise.

To start this week: create the staff register, write the parent paragraph, and choose one label for AI-assisted material. Send the paragraph in your next newsletter, and ask for questions. The first message tends to prompt more curiosity than concern, and a school that has already answered the obvious questions is in a better position for the harder ones. Revisit all three at your regular policy review.