Free resources are going live: 5 ready to download — free, no registration.Free resources →
BEETSAI

Safe-to-upload rehearsal: find and remove pupil information before using AI

Theme B · Responsible AI use · P4-P6 · Cross-curricular · Original idea #8

A hands-on practice for teachers: find and remove pupil-identifying information in six fictional school documents, then hand them safely to an AI tool. Includes an answer key, a preparation checklist and reusable steps.

Preparation
45 min (can be split across two lessons)
Equipment
Projector or computer; also runs on printouts alone
Version
0.1
Reviewer
Billy Kan
Fact-checked by
Billy Kan

What this solves

Teachers want to hand worksheets, reports or class photos to an AI tool for marking or organising, but these files often contain pupils’ names, student numbers, class, contact details, photos, or medical / special educational needs records. Once uploaded, this information can leave the school’s control.

This exercise does not lecture on abstract principles; instead it lets teachers practise once, hands-on: find the identifiable information in six fictional documents, remove it, and then judge what risks remain.

Learning objectives

By the end, teachers can:

  1. Identify the elements of school documents that count as “identifiable personal information” (not just names).
  2. Use the three techniques “remove → replace → blur” to turn a document into a version that is safe to upload.
  3. Judge the residual risks that remain after names have been removed (for example a combination of class + date + a special event).
  4. State the three checks to carry out before uploading.

Teaching steps

  1. Introduction (5 min) — Ask: “If you were going to hand a pupil’s composition to an AI for marking, what would you do first?” Collect the answers without judging them.
  2. Demonstration (5 min) — Project “Example 1” and demonstrate removing information step by step, explaining what is removed at each step and why.
  3. Group work (20 min) — Each group works on two examples (chosen from Examples 2–6) and fills in the “before-you-upload checklist”.
  4. Whole-class checking (10 min) — Check against the answer key in the attachments. Focus the discussion on “what risk remains after names are removed”.
  5. Wrap-up (5 min) — Together, write out a three-step checking prompt that the school can use.

Files and answer key

Six fictional examples, a printable checklist and the full answer key are in the download file:

Safe-to-upload check exercise (printable)

All examples are fictional and do not involve any real pupils.

Answer key — key ideas

  • A name is not the only identifier. Class + seat number, class + the date and nature of an event, and rare combinations (for example “the school’s only non-Chinese-speaking new arrival”), are all enough to identify an individual pupil.
  • Photos carry higher risk than text. A uniform, the classroom background, a name badge, or another pupil’s face can all amount to identification.
  • “Removed” does not mean “safe”. Ask: once the remaining pieces of information are combined, who else in the school could still recognise the pupil?
  • Marking does not require identity. If the AI is only being asked to mark, a pupil’s name is of no use to the AI and should always be removed.

Before-you-upload checklist (can be posted in the staff room)

□ Name, student number, class, seat number — removed or replaced with "Student A"
□ Contact details (phone, email, home address) — removed
□ Photo / video — removed, or cropped so no identifiable background remains
□ Special educational needs, medical, discipline or counselling records — removed
□ The combination of date + class + event — checked (the combination itself can identify)
□ File name / metadata — checked (file names often contain a name)
□ Do I really need to upload this file? — confirmed as needed for the purpose

How this relates to the EDB framework

This resource maps to G (the Guide to Using AI in Teaching in Primary and Secondary Schools, pp.2-4 and 2-9): the document requires learning data to be handled properly, and for identifiable parts to be removed before pupil work is uploaded for AI marking.

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

This page is a teaching resource independently produced by Beets and is not an official EDB document.

Feedback

Have you used this exercise? Which example was closest to your own situation? Tell us what needs adjusting, or suggest the next resource.

Source documents

  • G, pp.2-4 and 2-9

Official document: Blueprint for Digital Education Development in Primary and Secondary Schools

Related resources

Reuse

Download and adapt these resources freely for teaching; please keep the attribution. We collect no student work and nothing that could identify a student.