Do Canvas, Blackboard, and Moodle Each Store AI Detection Results Differently?

Yes — but only at the display layer: the AI report itself always lives on Turnitin’s servers, while Canvas, Blackboard, and Moodle differ in what they cache locally, where the score appears, and what survives course deletion or copying. Understanding how LMS store AI detection results turns out to matter in practical ways — for students wondering whether a flag follows them, for instructors hunting for last term’s reports, and for anyone involved in an academic integrity case trying to establish what the record actually is. Here’s the architecture, platform by platform.

Key takeaways

• One rule governs all three: the authoritative AI report is generated and stored by Turnitin. The LMS holds, at most, a cached number and a link.
• Canvas surfaces scores in SpeedGrader through the plagiarism framework — an indicator that opens Turnitin’s viewer. Canvas’s own database stores little more than the reference.
• Moodle‘s plagiarism plugin actively pulls score values into Moodle’s interface, so numbers display inline — the most locally-cached of the three.
• Blackboard links reports through the Grade Center / Gradebook into Turnitin’s viewer, closer to the Canvas model.
• Deleting or copying things in the LMS changes local references only. The Turnitin-side record persists under the institution’s account until a real deletion request removes it.

The architecture every platform shares

Start with what’s common, because it explains most of the confusion. None of the three LMS platforms performs AI detection. Each one forwards submissions to Turnitin (or another licensed service), and Turnitin generates the similarity report and — where the institution licenses it — the AI writing report. Those reports live in Turnitin’s infrastructure, attached to the submission record, under the institution’s Turnitin account.

What the LMS “stores,” then, is some combination of three things: the submission file itself (all three keep this — it’s ordinary coursework storage), a *reference* to the Turnitin record, and possibly a *cached copy of the score number* for display. The platforms differ in how much of the third item they keep and where they show it. That’s the whole difference — but it has real consequences.

Canvas: the linked indicator

Canvas integrates Turnitin through its plagiarism framework. When a student submits to an assignment with Turnitin enabled, the report status appears in SpeedGrader — a colored similarity flag, and for AI, the score visible when the instructor opens the Turnitin report viewer from the submission. Canvas itself is storing the submission plus the association to Turnitin’s record; the rich report — highlighted passages, the AI percentage with its caveats — renders inside Turnitin’s viewer, fetched live.

Practical consequences: if the Turnitin side is unreachable or the institution’s license lapses, old scores stop being viewable from Canvas even though nothing was “deleted.” And nothing about the AI result appears in the Canvas gradebook exports or APIs as first-class data — it’s Turnitin’s number, visiting.

Blackboard: through the Grade Center

Blackboard (Original and Ultra both, with interface differences) follows a similar linked model via the Turnitin LTI integration. Submissions flow to Turnitin; instructors reach reports through the assignment’s entry in the Grade Center or Ultra’s gradebook, which opens Turnitin’s report view. We’ve looked at how the AI score surfaces in Blackboard Ultra specifically — Ultra’s cleaner interface hides some of the plumbing, but the storage story is the same: Blackboard keeps the submission and the linkage; Turnitin keeps the reports.

One Blackboard-specific wrinkle: institutions migrating from Original to Ultra, or between Turnitin integration versions, sometimes orphan old report links. The reports still exist on Turnitin’s side and administrators can retrieve them — a detail that has decided more than one integrity dispute involving prior-term work.

Moodle: the score comes to you

Moodle is the outlier, and it’s the reason this question is interesting. Moodle core has no detection at all — why Moodle needs plugins for any detection at all is its own story — and the Turnitin plagiarism plugin fills the gap by *pulling data in*. The plugin submits work via Turnitin’s API, then fetches the resulting scores and writes them into Moodle’s interface: similarity percentages appear inline next to submissions in the assignment grading screen, with links out to the full viewer for detail.

That means Moodle’s database genuinely holds copies of score values — cached numbers the plugin retrieved. The full AI report with its highlighted passages still lives at Turnitin, but the headline figures are mirrored locally. Consequences: scores remain *visible* in Moodle’s grading screens even during Turnitin outages; the plugin’s data tables are part of Moodle backups in ways admins must handle deliberately; and site administrators have local database access to score data that in Canvas or Blackboard would exist only behind Turnitin’s API.

What survives deletion, copying, and time

The differences sharpen at the lifecycle edges:

• Deleting a submission or course in the LMS removes the local file and reference (and in Moodle, cached numbers). It does not touch Turnitin’s record. The paper, its similarity report, and its AI report persist under the institution’s Turnitin account until someone files an actual deletion request with Turnitin.
• Course copy duplicates structure, not submissions. No reports travel. Every platform behaves the same here.
• Retention is ultimately contractual — governed by the institution’s agreement with Turnitin — not by any LMS setting. Note the asymmetry within Turnitin’s side: repository storage (for future matching) is one thing; the AI report is an analysis attached to the submission record, not part of a matching corpus.

For students, the upshot is worth stating plainly: an AI flag doesn’t “follow you” between courses or platforms in any automated way — no future instructor sees a dossier — but the underlying records outlive the course shell, and institutions can retrieve them while their Turnitin account holds them. The record that matters most in any dispute remains the full Turnitin report, not the number an LMS cached.

And if what brought you here is worry about what those stored reports might say: the productive move happens before submission, not after. Check how your own writing reads while it’s still a draft — sentence-level results show exactly which passages would carry a flag. The free tier handles a full paper; pricing covers the rest, and there are more platform guides mapping each LMS in detail.

Frequently asked questions

Where does my Turnitin AI score actually live?

On Turnitin’s servers, in every case. Whether you submitted through Canvas, Blackboard, or Moodle, the AI report is generated and stored in Turnitin’s infrastructure, and the LMS shows it through a viewer or a cached number. The LMS differences concern what gets mirrored locally and how it’s displayed, not where the authoritative record sits.

Does deleting a course from the LMS delete the AI report?

Generally no. Removing a course, assignment, or submission from the LMS removes the local reference — the link, the cached score column — but the submission record and its reports persist on Turnitin’s side under the institution’s account. Actually removing a paper from Turnitin requires a separate deletion request through the institution.

Why does Moodle show a score column while Canvas makes me open a viewer?

Different integration architectures. Moodle’s plagiarism plugin fetches score values via API and writes them into Moodle’s own interface, so numbers appear inline in the grading screens. Canvas’s plagiarism framework surfaces the score in SpeedGrader as a linked indicator that opens Turnitin’s viewer for detail. Same underlying report, different amounts of locally cached display data.

Can a score visible in one course transfer when a course is copied?

No. Course copies duplicate structure — assignments, settings, due dates — not student submissions or their attached reports. AI and similarity results are bound to the original submission records. A copied course starts empty, and prior results remain accessible only through the original course or Turnitin’s own administrative interfaces.

Do all three platforms keep AI scores as long as similarity scores?

The retention clock is Turnitin’s either way, but the records differ: similarity matching may involve your paper being stored in the repository for future matching, while the AI report is an analysis attached to the submission rather than a matching corpus. Institutional contract terms govern how long submission records and reports remain retrievable on Turnitin’s side.