P4-S6 cross-subject: data-minimisation checklist — strip it before you type it

• StageP4-S6
• SubjectCross-curricular
• Learning strandreflection on the application of AI
• Official documentsGuide to Using AI in Teaching in Primary and Secondary Schools (Supplement II) · AI Literacy Learning Framework for Primary and Secondary Schools (Examples of Application Scenarios)
• Session time5 minutes the first time; about 30 seconds before each later input

One A4 page, four questions, a before-and-after comparison and a log: check each question before pasting, strip whatever answers yes, and type only when all four answer no.

• AudienceTeachers
• ThemeResponsible AI use

• EquipmentThe one-page A4 checklist (prints on its own); keep it beside the computer or the photocopier
• Version1.2
• Last updated2026-10-17

Deleting before you paste is easier than repairing after you paste. One page, four questions, checked in order: type only when all four answer no.

What this solves

Incidents rarely happen inside the tool. They happen at the moment of pasting. A teacher wants a set of exercises, a rewritten explanation, or a report built from a record; the whole document goes in, and with it a pupil’s name, class and seat number, a parent’s contact details, or a record that identifies exactly who it is about. Deleting the conversation inside the tool afterwards does not bring back the copy that has already left the school’s systems.

The official guidance is direct about this. The Guide to Using AI in Teaching requires schools using AI tools to collect and process student learning data to comply with the Personal Data (Privacy) Ordinance, the Copyright Ordinance and the related guidance, and requires a teacher who uploads pupil work to an AI platform for marking to remove the parts that identify a pupil first. In the Examples of Application Scenarios, the requirement is put as anonymisation: pupils entering a prompt or uploading material may supply anonymised data only, and must never enter, upload or store a real name, a student number or any other identifying information.

Both point at the same action: strip it before the input, not after. This page turns that action into four questions that can be answered one at a time.

What this page covers: the thirty seconds before the input. It does not repeat the list of what may and may not be entered (another resource sets that out item by item), it does not deal with parental consent (that belongs to another resource), and it is not a new school policy. Once all four questions answer no, anything further that your school requires is a matter for its existing procedures.

Learning objectives

Users will be able to:

1. Check a piece of text against four questions before pasting it.
2. Point out the identifying parts of a sample passage and write the stripped version.
3. Explain why deleting first is more reliable than repairing afterwards.
4. Record one check in the log: who entered it, when, the category of data, and whether it was anonymised.

How to use it

The checklist is one A4 page and prints on its own. The four questions run from easiest to hardest to judge: names and numbers first, then other details that identify a person, and only then whether a part is needed at all. Start at question one; if any question answers yes, strip that part and start again at question one. Type only when all four answer no.

Question 1 — Is there a name or a number that identifies a person directly?

A name, a student number, a class and seat number, a parent’s name or contact details, an identity document number. This is the easiest question to answer and the one most often half-done: the name is changed, and the class and seat number are left in place, which still identifies the pupil.

Question 2 — Is there any other detail that pins it to one person?

A face or a voice in a photograph, a recording or a video; handwriting; a school uniform or a classroom background; an incident only a few people know about. This is harder than question one, because it does not ask whether a name is written down but whether a reader or a listener could tell who it is.

Question 3 — Is this part actually needed for the task?

If removing it leaves the result the same, remove it. A whole document pasted in is usually half irrelevant to the task; asking why a section is there is more reliable than deleting it inside the tool later.

Question 4 — With it removed, can the tool still do what I need?

This is the hardest question, because it makes you think about the task itself. Strip too much and the tool cannot do the job; when that happens, the right move is to change the question (supply the task and the writing brief, for example, rather than the pupil’s own text), never to put the data back.

Before and after (all three rows are fictional examples, printed to show the format)

Log (one row per check)

The log is not there to assign blame. It does two things: it tells you what to strip next time you meet the same kind of material, and it lets the school say what was done if it is ever asked.

Answer key and marking notes

How to judge each question

What a good before-and-after looks like

• The stripped version still reads as coherent text, and the original task can still be done with it.
• The key linking a code to a real person stays in school and never appears in the input.
• Every item removed can be explained as unrelated to the task.

Common errors

• Changing only the name. Class, seat number, date and location together still identify the pupil.
• Pasting the key as well. The code then does nothing; the key stays in school.
• Pasting the whole document first. The data has already left the school’s systems; deleting the text inside the tool does not undo that.
• Stripping so much the task fails. That is what question four is for: change the question, do not restore the data.
• Reading anonymisation as “remove all context”. Anonymised means not identifiable, not meaningless.

What counts as meeting the standard here: all four questions answered, the sample rows stripped correctly, and one real row in the log. This is an operational tool, not something marked with a score.

Files

Data-minimisation checklist (editable Word document)

One A4 page: four questions with a yes/no box each, three before-and-after rows, and a three-row log. It prints on its own and can be kept beside the computer, the photocopier or in the staff room. Schools may reuse and adapt it.

Official documents

Every resource in this library is written to the AI Literacy Learning Framework for Primary and Secondary Schools and falls within the policy scope of the Blueprint for Digital Education Development in Primary and Secondary Schools.

This resource maps to:

• Guide to Using AI in Teaching in Primary and Secondary Schools (Supplement II) (pp.2-4 and 2-7): requires schools that use AI tools to collect and process student learning data to comply with the Personal Data (Privacy) Ordinance, the Copyright Ordinance and the related guidance, and requires a teacher who uploads pupil work to an AI platform for marking to remove the parts that identify a pupil first.
• AI Literacy Learning Framework for Primary and Secondary Schools (Examples of Application Scenarios) (p.17): states that pupils entering a prompt or uploading material may supply anonymised data only, and must never enter, upload or store a real name, a student number or any other identifying information.

The four questions on this page are this resource’s own classroom procedure, written from those requirements; they are not the wording of the documents, and the requirements above are what the documents themselves say.

Learning strand: reflection on the application of AI

Official documents page: https://www.edb.gov.hk/en/edu-system/primary-secondary/applicable-to-primary-secondary/it-in-edu/debp.html

This page is an independent Beets teaching resource, not an official EDB document.

Feedback

If you have any comments or questions about this resource, or an idea for the next one, tell us on the feedback page.