Don't Feed the Machine Your Students' Names

The most common student privacy violation involving AI doesn't look like a hack. It looks like a tired teacher at 9 p.m., pasting a paragraph into a free chatbot: "Help me rewrite this feedback for Maya, she's struggling with..."

Nothing malicious happened. But a student's name, tied to a description of her struggles, just left the school and entered a system the school doesn't control. Multiply that by a thousand small moments across a district and you have a data practice nobody approved and nobody can audit.

Here's the habit that fixes it: strip identity before you prompt. "Help me rewrite feedback for a 7th grader struggling with thesis statements" gets you the same help with none of the exposure. The AI doesn't need Maya's name to improve your sentence.

This is the same discipline I learned in my federal grants compliance days — we never moved identifiable data further than it needed to go. Data minimization isn't a regulation buzzword. It's the practice of asking, before every paste: does the tool need to know who this is?

Teach this one habit to every teacher in your building and you've done more for student privacy than most software purchases will.

I explain frameworks; I don't give legal counsel.