What AI detectors do, in plain terms
An AI detector is software that looks at a piece of text and gives a score or label suggesting how likely it is that AI wrote it. It does not see who sat at the keyboard or how the work was made. It only guesses from patterns in the words.
Because it is a guess, it can be wrong in both directions: a false accusation of a student who wrote honestly, or a miss on text that AI wrote or heavily shaped.
Why a score is not proof
Student writing is often simple, careful and formulaic, especially when students are following a structure a teacher taught. That kind of writing can resemble what a tool flags. Light AI help, such as a grammar fix, or edits made after AI drafting can also change how a tool responds.
A score gives no explanation a teacher can check, and a student cannot easily rebut a number. For these reasons, a score should not be treated as evidence on its own.
Fairness risks worth considering
Any tool that judges writing by its patterns carries a risk of treating some students unfairly. It is worth asking, before relying on one, whether writers who use a more formal or limited vocabulary could be flagged more often. This could include students who are writing in a second language or students who follow a template closely. This is a risk to weigh carefully, not a settled finding, and it is one more reason to avoid using a score as a verdict.
Student privacy and uploading work
Pasting or uploading a student's work into an online tool means sending it to a third party. Before doing so, check your board's rules, your province or territory's privacy legislation, and the tool's current terms. If you are unsure, do not upload. See also What student information should never go into an AI tool?
Better alternatives
Evidence of the student's own process is stronger than any score:
- Outlines, drafts and revision history.
- A short log of how AI was used, when permitted.
- Source checks against real, non-AI sources.
- A brief conversation in which the student explains choices in the work.
See Can teachers tell if a student used AI? and How do you redesign assignments for the age of AI?
If your school does use a detector
Follow your board's policy and any guidance on approved tools. Treat the result as a prompt for a conversation, not a verdict. Never base a grade or a consequence on the score alone, and be ready to show other evidence. If a concern holds up, the steps in What should a teacher do when a student used AI when they were not allowed to? apply. The wider context is in AI and academic integrity.
Quick answers
Can a detector prove a student used AI?
No. A detector gives an estimate, not proof, and it can be wrong in either direction. Use it, if at all, only as a reason to start a conversation.
Should I paste student work into a detector?
Only if your board allows it. Student work is personal information, so check your board's rules, your province or territory's privacy legislation and the tool's terms first.