Updated August 2026 · read from Turnitin and Instructure docs
Canvas AI detector: does Canvas have AI detection, and can Canvas detect AI?
Canvas has no AI detector. It is a learning management system, and it never reads your writing. Everything people call "Canvas AI detection" is a separate product bolted onto it.
Below: what actually does the scanning, and the four conditions that all have to be true before one sentence of your work is scored.
2 free runs a day, up to 200 words each. We save your run so you can get back to it, and a delete button appears with the result. See privacy.
This is our own AI-pattern score, measured here on sentence rhythm, template phrases, vocabulary variety and passive voice. It is not a GPTZero, Turnitin, Originality.ai, Copyleaks or ZeroGPT result, and it does not predict one. Worth knowing: we also ask the rewrite to vary sentence length, drop template phrases and prefer the active voice, so some of the drop is built in. Read the two panels below, not just the number.
Before ·
After ·
Your own text rewritten · Saved to your history, delete any time
The short answer
There is no Canvas AI detector. Instructure's Canvas does not analyze student writing and shows no AI score anywhere in its interface. Turnitin, whose detector powers most of the AI scoring that happens in United States higher education, says so directly: "There is no AI writing indicator or score embedded directly in the LMS user interface and users will need to go into the report to see the AI score." When a Canvas submission is checked for AI, four things had to line up first. The school bought Turnitin. The school licensed AI writing detection, which "is only available to customers that license Turnitin Originality". An administrator left the feature enabled. And your instructor ticked Turnitin on that individual assignment, because in Canvas it is a per-assignment setting rather than a course-wide or campus-wide one. Break any of the four and the submission is never scored.
Last updated August 16, 2026. Every quoted line was read from Turnitin's and Instructure's own documentation on that date.
The conditional answer
The four gates between a Canvas submission and an AI score
Pages answering this question tend to describe Canvas AI detection as a switch that is either on or off for a whole university. In Canvas it is four separate decisions made by four different parties, and the last one is made assignment by assignment.
| Condition | Decided by | What the documentation says | If it is not met |
|---|---|---|---|
| 1. Canvas itself runs no detector | Instructure | Canvas is a learning management system. It stores and routes submissions, it does not analyze writing. Turnitin puts this plainly: "There is no AI writing indicator or score embedded directly in the LMS user interface and users will need to go into the report to see the AI score." | Every AI score you have ever heard of in Canvas came from a separate product plugged into it. |
| 2. Your institution licenses AI detection | Your school | AI writing detection "is only available to customers that license Turnitin Originality". A university can run Turnitin Similarity or Feedback Studio inside Canvas for years and never have the AI feature at all. | The Similarity Report your instructor opens has no AI indicator in it. Nothing is scored. |
| 3. An administrator leaves it switched on | Your admin | Licensing is not the same as running it. Admins "have the option to enable/disable the AI writing feature from their admin settings page", and Turnitin "will only process submissions for AI writing detection if the institution has the feature enabled". | Submissions pass through Turnitin for similarity only. No AI percentage is ever calculated. |
| 4. Your instructor enabled it on that specific assignment | Your instructor | This is the gate unique to Canvas. Turnitin has to be turned on per assignment. In the Plagiarism Framework workflow the submission type must be Online Submission with Text Entry or File Uploads, and only then does "the option to enable Turnitin for the assignment show up in the Plagiarism Review box". In the LTI workflow the instructor has to build a Turnitin external-tool assignment instead. | A plain Canvas assignment with the box unticked never reaches Turnitin. It is graded in SpeedGrader and that is the end of it. |
The fourth gate is the one specific to Canvas, and it is the reason two assignments in the same course can be treated completely differently. Turnitin is not a campus-wide filter that every Canvas submission passes through. In the Plagiarism Framework workflow, an instructor building an assignment has to set the submission type to Online Submission with Text Entry or File Uploads, and only then does the Turnitin option appear in the Plagiarism Review box for them to tick. In the LTI workflow they have to deliberately create a Turnitin external-tool assignment instead of a normal one. An instructor who does neither has built an assignment that no detector will ever see.
The first three gates are Turnitin's, and they apply wherever Turnitin runs. We take them apart in full, with the licensing and administrator quotes in context, in whether Turnitin checks for AI at all. The short version is that an institution can be a Turnitin customer of fifteen years' standing and have no AI detection anywhere in its workflow, because AI writing detection arrived later as part of Turnitin Originality and has to be bought separately.
None of this is visible to you. The AI indicator is hidden from students whether it is enabled or disabled, so a course with detection running and a course with none look identical from the submission page. That is worth sitting with before you go looking for a way to check.
Coverage
What in Canvas gets checked for AI, and what never does
Even at a school with everything enabled, most of what you submit through Canvas is not eligible for an AI reading. The submission type and the word count decide it before any model runs.
| What you submit in Canvas | Checked for AI? | Why |
|---|---|---|
| File upload assignment with Turnitin enabled | Yes | The standard path. The file goes to Turnitin, comes back with a Similarity Report, and the AI indicator sits inside that report if the first three gates are met. |
| Text entry assignment with Turnitin enabled | Yes | Text Entry is one of the two submission types that make the Plagiarism Review option appear, so typed-in answers are processed the same way a file is. |
| Assignment with Turnitin not enabled | No | Nothing is sent anywhere. This is a per-assignment choice the instructor makes when building it, and plenty of assignments never have it ticked. |
| Discussion board posts | No | Discussions are not assignments with a submission type of Online Submission, so the Plagiarism Review box does not apply to them. Most posts also fall well under Turnitin's 300-word floor. |
| Anything under 300 words | No | Turnitin treats 300 words of prose as a file requirement rather than a preference. Short reflections and weekly posts come back with no AI reading at all. |
| Quizzes and quiz answers | No | Quiz submissions do not go through the assignment plagiarism workflow. Canvas records a quiz activity log, which is a list of page events, not a writing analysis. |
| Work submitted before the feature was turned on | No | Turnitin: "we cannot retroactively process submissions. If you would like to process past submissions for AI writing, you will need to re-submit the document." |
| Tables, bullet lists, code, bibliographies | No | The model reads "individual sentences contained in paragraphs that make up a longer piece of written work". Everything else is stripped out before the percentage is calculated. |
The discussion board line surprises people, because weekly posts feel like the most AI-drafted writing on any course. The reason they escape is structural rather than generous. Turnitin's Canvas integration attaches to assignments, and specifically to assignments whose submission type is Online Submission with Text Entry or File Uploads. A discussion is a different object in Canvas. On top of that, Turnitin requires at least 300 words of continuous prose before it will return a score, and a typical forum reply is a third of that.
Turnitin also strips out everything that is not prose before it calculates anything. Its qualifying text is "individual sentences contained in paragraphs that make up a longer piece of written work", which leaves out tables, bullet lists, code and bibliographies. The consequence is one most students get wrong: the percentage "does not necessarily correlate to the amount of text in the submission". A lab report that is half tables can return a figure drawn from a much smaller body of writing than its page count suggests.
The number you can actually see
That colored percentage in Canvas is not an AI score
This is the single most common misreading of Canvas, and it costs students a lot of unnecessary panic. Turnitin's Canvas documentation says that "indications of the Similarity Report percentage will appear in the Canvas Gradebook". Similarity. That number is the text-matching score, the old plagiarism measurement that compares your writing against web pages, publications and a repository of student papers. It has nothing to do with AI.
The AI percentage lives somewhere else entirely, and Turnitin is explicit that the two are "completely independent and do not influence each other". You can hand in a wholly AI-written essay on an obscure topic and see a similarity score near zero, because nothing you wrote matches anything already published. The reverse is just as true: a heavily quoted, properly cited literature review can show a high similarity figure while the AI reading is nothing at all.
You will not see the AI figure in any case. Turnitin states that "only instructors and administrators are able to see the indicator" and that "the AI writing detection indicator and report are not visible to students". Instructors can download the AI report as a PDF and share it if they choose, which some do and most do not. So if a percentage is visible to you inside Canvas, it is a similarity score by definition, and reading it as an AI score is a mistake.
Instructors also control when you see even that. Turnitin's Canvas settings let them release the Similarity Report immediately, after grading, after the due date, or never. And a second matching pass, the collusion check that compares classmates against each other, runs on the assignment due date rather than at submission, which is why a score sometimes changes after you thought the process was finished.
How detection reaches Canvas
The three ways Turnitin plugs into Canvas
Turnitin publishes three Canvas integrations, and they do not behave the same way. Which one your course uses changes where the report opens and what your instructor can do with it.
Turnitin LTI 1.3
Recommended by Turnitin for new setups
Runs as a Canvas External Tool with its own assignment inbox. Grading happens in Turnitin Feedback Studio, and Turnitin notes that Canvas SpeedGrader annotation and grading tools cannot be used with it. If your school runs Canvas Open Source, Turnitin says the LTI is the only integration available.
Turnitin LTI 1.1
Still in use, not recommended for new users
The older external-tool integration. Similarity Reports launch from both the Turnitin inbox and Canvas SpeedGrader and are indicated in the Canvas Gradebook. Turnitin points existing users toward LTI 1.3 as the industry standard.
Canvas Plagiarism Framework
Built into the normal Canvas assignment
The workflow most students actually meet. The instructor builds an ordinary Canvas assignment and ticks Turnitin in the Plagiarism Review box. Turnitin describes it as designed "to provide a similarity report and spelling and grammar suggestions only", with all feedback and grading done in SpeedGrader.
Instructors can run both workflows in the same course, using the LTI for one assignment and the Plagiarism Framework for another, so the presence of Turnitin on your last essay tells you nothing about your next one. The practical tell, if you want one, is where the report opens. A Turnitin assignment that launches its own inbox is the LTI. An ordinary-looking Canvas assignment where a similarity figure appears in the Gradebook is the Plagiarism Framework.
Turnitin is not the only option a school can connect. Copyleaks and GPTZero both integrate with Canvas through LTI, and Instructure has published a case study of a community college running GPTZero across its Canvas environment. Turnitin is the one with the deepest Canvas documentation and the widest footprint in United States higher education, but "what AI detector does Canvas use" has no single answer, because Canvas does not choose. Your provost's office did. We compare how the major detectors actually perform in which AI detector is most accurate.
The other half of the question
What Canvas does record, and how much it is worth
Saying Canvas has no AI detector is not the same as saying Canvas sees nothing. It logs submission timestamps, page views, and for classic quizzes an activity log of events like starting a question or leaving the quiz page. Instructors can open that log. Plenty of forum advice treats it as a surveillance system, and it is not.
Instructure sets the expectation itself. Its guide on viewing a student's quiz log states that quiz logs "are not intended to validate academic integrity or identify cheating for a quiz", and that the feature "is designed to help you investigate problems that a student may have in the quiz and gain insight into your quiz questions". Instructors who have tested the log report it firing when a cursor drifts to the browser chrome and staying silent when a student genuinely switches tabs, which is what you would expect from an event logger that was never built as a proctor.
More to the point for this question: none of it is writing analysis. A timestamp says when you submitted, not who wrote it. There is no keystroke recorder in the Canvas submission box, so pasting a finished draft into a text entry field is not itself an event Canvas flags. Tools that do record drafting behavior, like the authorship features built into some word processors, work inside the editor you drafted in rather than inside Canvas.
That distinction between guessing from finished text and recording how a document was written matters more each year, and it cuts both ways. Detection tools infer. Provenance tools observe. If you are ever on the wrong end of a false flag, a version history showing an essay developing across days is worth more than any argument about detector accuracy, which is the case we make in how to prove you did not use AI.
The practical part
How to work out what your own course does
Look at the assignment, not at Canvas. This is the one thing you can genuinely check. If an assignment launches a Turnitin inbox or shows a similarity figure after you submit, Turnitin is attached to it. If it behaves like a plain Canvas upload with nothing but a grade coming back, the fourth gate was never opened for that assignment. This tells you about similarity checking rather than AI scoring, but it is the only observable signal you have.
Read the integrity policy instead of the forums. Whether your institution licensed AI detection and left it enabled is written down somewhere, usually in the academic integrity policy or the library's guidance. A Reddit thread about a different university's Canvas instance tells you nothing, because these are four local decisions.
Keep the drafting history. Write in something that records revisions, and avoid the pattern of pasting a finished draft into an empty document. This costs you nothing and is the only evidence that speaks to the actual question if a score is ever queried.
Assume it is on. Since the indicator is invisible to you either way, the cost of guessing wrong is asymmetric. Writing as though every submission is scored is the cheap option, and it does not mean gaming a percentage. Turnitin publishes the prose most likely to be misread by its own model, and the categories are telling: "content without a lot of structural variation, text that literally repeats itself, or text that has been paraphrased without developing new ideas".
That last category is the one worth understanding, because it is where most rewriting advice goes wrong. Swapping synonyms keeps the uniform rhythm and the recycled structure that a detector responds to, so a lightly reworded draft frequently scores worse rather than better. Our humanizer works on the level that matters instead, rebuilding sentence length, structure and register so the prose varies the way a person's writing varies. It cannot tell you whether your school has detection enabled, and it will not promise you a number, because Turnitin has deliberately made both of those unknowable.
Frequently asked
Canvas AI detection questions, answered
Does Canvas detect AI?
Canvas does not detect AI by itself. It has no writing detector of its own and displays no AI score in its interface. AI detection reaches a Canvas submission only when a third-party product, almost always Turnitin, has been licensed by the school, enabled by an administrator and switched on by your instructor for that assignment.
Does Canvas have AI detection?
Not as a built-in feature. Turnitin states that "there is no AI writing indicator or score embedded directly in the LMS user interface". Where a school has AI detection, the score lives inside the Turnitin Similarity Report, which an instructor opens through Canvas. Canvas is the door, not the detector.
Can Canvas detect AI?
Only in the sense that a hallway can detect a fire, by having an alarm mounted in it. The analysis is done by Turnitin or another integrated tool, and it only happens on assignments where that tool was enabled. An assignment with no plagiarism integration ticked is never checked for anything.
What AI detector does Canvas use?
Canvas does not use one. Schools choose. Turnitin is by far the most common integration in United States higher education and is the one with published Canvas documentation, but Copyleaks and GPTZero also connect to Canvas through LTI. Which one your school runs, if any, is a local decision.
Does Canvas use Turnitin?
Only if your institution buys Turnitin and connects it. Turnitin publishes three Canvas integrations: LTI 1.3, the older LTI 1.1, and the Canvas Plagiarism Framework that works inside the standard Canvas assignment. None of them are present by default, and none of them are on until someone turns them on.
Does Canvas detect AI in discussion posts?
Almost never. The Turnitin option only appears on assignments whose submission type is Online Submission with Text Entry or File Uploads, and discussions are not built that way. Even where a graded discussion did reach Turnitin, most posts sit under the 300-word minimum the detector requires.
Does Canvas detect AI without Turnitin?
Not on its own. If your school has no plagiarism integration, Canvas records what it always records: submission timestamps, page views and quiz activity. That is metadata about when you did something, not analysis of what you wrote, and it produces no AI score.
Can Canvas detect ChatGPT?
Canvas cannot. Turnitin can, where it is enabled, and it lists the GPT-5 family, GPT-4o, Claude, Gemini, Grok, LLaMA, Mistral, DeepSeek and others among the models it detects. It works from the finished text alone. It has no view of your browser, your ChatGPT account or your prompts.
Can Canvas detect copy and paste?
Canvas does not flag pasting into an assignment. It has no keystroke recorder in the submission box. Some provenance tools built into word processors do record pasting, but that is a feature of the tool you drafted in rather than anything Canvas observes.
Does Canvas detect Grammarly as AI?
Canvas detects nothing, so the real question is whether Turnitin does. Turnitin scores the finished text and cannot see which browser extensions touched it. Ordinary grammar and spelling corrections leave your sentence structure intact, which is what the model reads. Rewriting whole sentences is a different matter.
Can Canvas detect AI extensions during a quiz?
Not by inspecting your browser. Canvas keeps a quiz activity log that records events like leaving the quiz page, and Instructure warns that quiz logs "are not intended to validate academic integrity or identify cheating for a quiz". Detecting an installed extension is outside what Canvas can see.
Does Canvas automatically detect AI?
No. Nothing about AI detection in Canvas is automatic from a student's point of view. Four separate decisions have to line up first: Turnitin bought, AI detection licensed, the admin toggle left on, and the instructor ticking Turnitin on that particular assignment.
Keep reading
The detection questions behind this one
Does Turnitin check for AI?
The three licensing and administrator conditions behind every AI score, quoted from Turnitin's own FAQ.
Can Canvas detect ChatGPT?
What travels from a Canvas assignment to Turnitin, and which models Turnitin says it recognizes.
Turnitin AI checker
Why no public Turnitin AI checker exists, and what the sites offering one actually are.
How accurate is the Turnitin AI detector?
Every accuracy figure traced to its source, including the several with no traceable source at all.
AI detector false positive rate
The peer-reviewed measurements, including the study where detectors beat human graders.
Best AI humanizer for Turnitin
What rewriting can realistically change before a submission, and what it cannot touch.
Write drafts that read like a person wrote them
You cannot see which of the four gates your course has open, so the only part worth controlling is the writing itself. Our humanizer takes the flat, uniform register out of an AI draft so it reads the way you do.
Plans planned from $9/month billed yearly · Your runs saved to your history, delete any time