Ethics of teaching
~ min read
30-second summary
- Three axes where the teacher’s responsibility doesn’t shift: honesty about how you use AI, privacy of minors’ data, grading that you sign.
- Honesty: routine prep needs no disclosure, but you check what you give students before you give it. And don’t hold them to a standard of honesty about AI you don’t meet yourself.
- Privacy: minors’ data is among the most sensitive. No names, grades, diagnoses, or class photos in a consumer chat. The gradebook doesn’t go into the chat.
- Grading: the AI proposes, you decide and sign. A grade you didn’t verify is still yours, and you answer to the family for it.
- Institutional policies and guidance change: before you adopt a practice, check your own school’s policy and talk to your data-protection lead.
The module opened with a contract: AI belongs in preparation, it doesn’t take your place in class. Eight lessons later, you’ve seen how to use it to design, explain, differentiate, include, build assessments, grade, communicate with families, and recognize it when it ends up in the students’ hands. What’s left is the question that holds the rest together: when AI enters your work, which responsibilities stay yours and don’t get delegated?
The answer isn’t a regulation to memorize. It’s three axes where your responsibility doesn’t shift, whatever the AI did. The thread tying them together is the same one from Things NOT to do: who signs is who answers. At the front of the class, the one who signs is you.
Honesty: about how you use AI
Section titled “Honesty: about how you use AI”Do you have to tell students that you prepared the worksheet with AI? For routine prep, no: you don’t disclose which textbook you prepared the lesson from or which website you took a diagram from, and AI is just one more prep tool, like any other. But three things change the picture.
First: you check what you give students before you give it. The AI can put a wrong date in a worksheet, an imprecise step in an explanation, a mistake in an answer key. If you hand out the material without rereading it, you’re teaching the mistake, and the responsibility is yours, not the machine’s. Preparing with AI doesn’t remove the teacher’s check, it makes it faster.
Second: don’t hold students to a standard of honesty about AI you don’t meet yourself. If you forbid them from having the AI write their essays but you generate your feedback in bulk without rereading it, the contradiction shows, and it undermines your credibility on that score. The best way to teach honest use of the tool is to model it: the AI helps, but the thinking and the signature stay with the person.
Third: with families, AI is never a source of authority. A line like “The artificial intelligence suggests that your child…” in a communication is an institutional own goal. The communication is yours, you sign it, you stand behind it (the pattern is the one from Emails to families).
Privacy: the data belongs to minors
Section titled “Privacy: the data belongs to minors”A minor student’s data is among the most sensitive you handle: name, grades, a learning-disability diagnosis, a family situation, a behavior note. The model you saw in What you share when you use AI and in Company data and privacy applies here with even more force: not because the rule changes, but because the subject is a minor and GDPR grants them stronger protection: a minor can’t fully weigh the consequences of sharing their own data.
In practice: no student names, no grades tied to people, no diagnoses, no family details go into the chat. You anonymize first, always, as in grading and in emails. The electronic gradebook is the trickiest case: it holds grades tied to names, that is, minors’ personal data at scale. You don’t paste it into a consumer chat to get it to “work out a few averages”. If you need a calculation, pull the numbers without the names, or use the gradebook’s own tools. The AI feature now built into many gradebooks is a different case: there the provider is a data processor under contract with the school, so it falls under that agreement, not your personal account. Before using it, check that the contract covers it and that the data isn’t used for training.
You don’t carry this framework alone. Almost every institution has a data-protection lead (often a Data Protection Officer, DPO) and a policy on the use of digital tools. Public schools are legally required to have one: if you don’t know who it is, ask the front office or the head teacher. A small private body without one still follows the principle, and you turn to whoever handles data and contracts. Before you bring a new practice into class, you check that policy: it’s there to protect the students and you too.
Grading: the signature stays yours
Section titled “Grading: the signature stays yours”The grade is an act you’re accountable for, to the student and the family. The AI can propose a correction, apply a rubric, flag where a paper departs from an answer key, but it doesn’t issue grades. The pattern is the one from Grading with criteria: the AI proposes, you decide. A number you didn’t verify doesn’t stop being yours just because a machine suggested it.
This holds even more sharply in the collective moments. The end-of-term grading meeting is a human deliberation, made of people who know the student, who weigh a path and not just an average. The AI has no seat at that table. It can help you organize the numbers beforehand, like a smart spreadsheet, but the assessment you bring to the meeting is yours, and it stays a judgment, not an output.
What NOT to do
Section titled “What NOT to do”Don’t hand out AI material without rereading it. A mistake you give to the class becomes a mistake you teach, and the responsibility is yours.
Don’t pour minors’ data into a consumer chat. Names, grades, diagnoses, photos: you anonymize first, or you use the institutional tools. No convenience justifies the exception.
Don’t delegate the grade. The AI without your check isn’t a second marker, it’s a generator of plausible numbers. The decision, and the signature, remain yours.
Check what you understood
Section titled “Check what you understood”What comes next
Section titled “What comes next”With this lesson the For teachers module is complete: you have a method to design, teach, assess, and communicate with AI alongside you, without handing it the piece of work that stays yours. Come back to these lessons when you need them.
From here the handbook goes a level deeper. For anyone who wants to understand how AI does what it does, all the way to building their own tools on top of it, Module 6 Going deeper starts here and goes under the hood: APIs, context in tokens, RAG, agents. It’s the most technical audience of the handbook, but the thread is the same one you’ve followed so far: know what the tool does so you can use it with judgment.