Turnitin flagged a submission in Moodle. What happens next?

A detector puts a number next to a submission. Turnitin’s AI indicator says 78%, or the similarity report says 41%. In Moodle, that is where the software stops. Everything after it, looking into it, asking the student, deciding, handling the appeal, is up to people, and in most schools it runs on email.

That is a problem, because every serious source says the number is the start of a process, not the end of one.

Is a high AI score enough to accuse a student?

No. The University of Melbourne’s guidance for staff is blunt: “A high AI detection score alone does not constitute grounds for making an allegation of academic misconduct.” Staff are told to find other evidence first.¹1 University of Melbourne, Turnitin’s AI writing detection tool, staff guidance, checked 4 November 2026.

The detector vendors say the same about their own tools. Originality.ai and Copyleaks both describe their results as something to review, not a finding of misconduct.²2 Originality.ai, Moodle plugin; Cochise College, Copyleaks plagiarism and AI detection software. Detectors misread human writing, and AI text can be edited until it passes. The score tells you where to look.

What does a fair process look like?

The details vary by institution, but the shape is the same almost everywhere:

1. Report. Someone notices something and writes down what, with the evidence: the report, the essay, earlier work to compare.
2. Investigate privately. Before the student hears anything, staff check whether there is a case. Many flags end here.
3. Hear the student. The student is told what was found and asked for their side: drafts, notes, how they wrote it.
4. Decide. Someone records an outcome and the reasons. Common outcomes run from no case to answer, through an educational warning or a resubmission, to a reduced mark or a referral to a panel.
5. Appeal. The student can challenge the decision, usually within a fixed window. The University of Galway, for example, gives fourteen days from the decision email and accepts appeals on grounds such as procedural errors or new evidence.³3 University of Galway, How to appeal an academic integrity decision.
6. Keep the record. Who did what, when, and what the student was told. When a decision is challenged, the record is what you defend it with.

What does Moodle give you for this?

Very little. Moodle has the submission, the detector’s score (through the Turnitin or other plagiarism plugin) and the grade. It has nowhere to put a case: no stages, no private notes, no place for the student’s response, no record of the decision.

When we searched the Moodle Marketplace in November 2026, we found detectors, proctoring tools, and a plugin that shows an integrity statement students click to accept. Nothing that tracks the case itself.⁴4 Moodle Marketplace searches for “integrity”, “misconduct” and “plagiarism case”, 4 November 2026. So it ends up in inboxes and spreadsheets, where three problems keep coming back:

• Students hear too early, or too late. Without a clear “investigating” stage, a student can be emailed about a flag nobody has checked, or not told until a decision is made.
• Staff notes leak. A remark meant for a colleague ends up in a thread the student sees.
• No record at appeal time. Months later, nobody can show what the student was asked, when, and what they answered.

How we handle it in Moodle

We built the missing part as an open source plugin: Academic integrity cases. It adds a case inside the course, about one student and optionally one activity:

The server enforces the order: a case can’t be decided before someone investigates, and only the student can appeal. Private staff notes and evidence files never reach the student. Every action is recorded with who did it and when. The student is notified through Moodle’s own messaging, and the person handling the case hears when the student responds or appeals.

It also covers the parts universities get audited on: Moodle’s privacy API for data requests, log events, and a report builder source so a quality office can count cases by course and outcome.

It works next to any detector. Use Turnitin, Originality.ai, Copyleaks or nothing at all: a teacher’s own judgement can start a case too.

Can it follow our policy?

Out of the box it follows the six stages above, which match most policies we’ve read. Where yours differs (a panel that decides, deadlines for the student’s response, a link from the grading page, a different set of outcomes), that’s a few days of work on top of the plugin. Our prices for a change like that start small, and you can get a fixed quote in a few minutes.

The plugin is free under the GPL, like Moodle. The code and the test reports are on GitHub.