There's a meeting that happens in schools now, thousands of times a term, and almost everyone walks out of it worse off.
A teacher reads an assignment that doesn't sound like the student. Maybe a detector tool put a scary percentage on it. The student gets called in. The teacher says some version of "this doesn't feel like your work." The student says they wrote it. And then... nothing. There's nowhere to go. It's a feeling against a denial, with a grade and a relationship hanging on whoever blinks first.
If the student was honest, you've just told a hardworking kid you don't believe them, and they will remember that longer than anything you taught them. If the student wasn't honest, they learned that denial works. Both outcomes are corrosive, and the detector percentage doesn't rescue either one, because the error rates on those tools are well documented and every student knows it.
The problem isn't that teachers are bad at these conversations. The problem is they've been sent into them with no evidence.
What a receipt looks like
Now run the same meeting with one change. Before anything was ever submitted, the school's devices were checking AI prompts as students typed them. On Tuesday at 2:14pm, on lab machine 17, someone typed "write my whole history essay on the Treaty of Versailles, 900 words" into a chatbot, and the system blocked it and kept the context: the host, the tool, the time, the prompt itself.
The conversation is unrecognizable. There's no accusation built on writing style, no percentage, no "it doesn't feel like yours." There's a specific moment to talk about. When did this happen, what was going on, what were you stuck on, why did the essay feel impossible at 2pm on a Tuesday?
Teachers who have run both versions of this meeting tell us the second one is not even the same genre. The first is a trial. The second is teaching.
The quiet upside: protecting honest students
The part I find most underrated: receipts protect the innocent more than they catch the guilty.
When a strong student turns in strong work, and a teacher's only tool is suspicion, strong work itself becomes suspicious. That's the trap of the detector era. But when the school's actual record shows no blocked attempts, no flagged prompts, nothing, the strong work just gets to be strong work. The kid who wrote a great essay gets a compliment instead of an interrogation.
Visibility, done right, is an exoneration machine.
Done right is doing a lot of work in that sentence
To be clear about what this isn't: it isn't reading students' messages, scanning their documents, or surveilling their browsing. The check happens at one narrow point, the moment a prompt is headed to an AI tool on a school-managed device, against rules the school wrote in plain language. Clear violations get blocked and recorded. Everything else passes through untouched and unlogged.
Narrow scope is what makes the whole thing defensible in front of parents, and being defensible in front of parents is the actual bar for any school technology. A dashboard you'd hesitate to demo at a parent evening is a dashboard you shouldn't deploy.
The relationship is the point
Academic integrity work was never really about catching people. It's about keeping the diploma honest and keeping trust between teachers and students intact while doing it. Suspicion erodes that trust from both directions. Evidence, scoped tightly and shown honestly, protects it.
Have the conversation with a receipt in hand. Better yet, build the system where most of those conversations never need to happen, because the dishonest prompt got a polite no before the essay ever existed.