Does Canvas have AI detection? What instructors can and can't see (2026)

No, Canvas doesn't have AI detection built in. Instructure's own staff said so plainly on the Canvas Community forum: "At this time, Canvas does not provide instructors with access to an AI generated paper detection feature." What your instructor can see depends on what your school has plugged into Canvas, usually Turnitin, sometimes GPTZero or Copyleaks, plus the quiz logs Canvas keeps by default.

That's the short version. The longer one matters more, because "Canvas" on a student's screen often means three or four separate products stitched together, and each one sees something different. I've pieced this together from Instructure's guides, university IT pages and the detector vendors' own documents, as of late October 2026. Nobody here ran submissions through any of these tools; this is a research piece, and it won't tell anyone how to dodge a checker.

Does Canvas have AI detection? The quick answer

Canvas itself doesn't scan essays for AI writing. Even GPTZero, which sells a Canvas add-on, admits it on its integration page: "Canvas does not currently include AI detection".

So when people ask "what AI detector does Canvas use?", the honest reply is: none, unless your institution bought one. Schools add detectors as LTI apps or through the Canvas Plagiarism Framework, and the results then show up inside SpeedGrader. To a student, that looks like Canvas checking for AI. It isn't, strictly speaking (a small distinction, but it changes who you'd appeal to if a score looks wrong).

Canvas overview: the AI features Instructure actually ships

Instructure has been adding AI to Canvas under the IgniteAI name. The December 2025 release announcement lists a Rubric Generator, Discussion Insights (which flags engagement trends and "students who may need additional support") and IgniteAI Translations for discussions, inbox messages and announcements. None of those is a detector.

At InstructureCon 2026 in Louisville, the focus stayed on teaching tools. Wake Forest's conference recap mentions the IgniteAI Agent, an AI Readiness Certification, "course level controls" so faculty can define acceptable AI use, and three new tiers: Canvas Core, Canvas Plus and Canvas Next. I'd read that as Instructure picking a side: it wants to help instructors set AI rules, and it's leaving the policing to partners. That seems sensible to me, given how shaky detection is.

Key features instructors can use

Plagiarism Review through the Canvas Plagiarism Framework

This is the main door detectors come through. When an instructor creates an online assignment, there's a Plagiarism Review drop-down. The University of Minnesota's setup guide walks through picking "Turnitin-Plagiarism Framework" there, choosing repository and comparison settings, then opening the similarity percentage from SpeedGrader. Students can see the report too, if the instructor allows it.

Copyleaks hooks in the same way. Azusa Pacific's instructor guide has faculty select the Copyleaks tool in the same Plagiarism Review section, and lets them show the report to students immediately, after grading, after the due date or never.

Turnitin as an LTI 1.3 tool

Some schools run Turnitin as a separate LTI 1.3 app instead. Going by UMN's instructions, those submissions aren't shown in SpeedGrader at all; instructors grade inside Turnitin Feedback Studio, and Canvas rubrics can't be used. The same page adds a detail that surprises a lot of students: "AI Writing Indicator is not enabled for the University of Minnesota." Turnitin being in your course doesn't mean its AI checker is switched on. Schools choose.

GPTZero for Canvas

GPTZero sells a per-institution Canvas integration. On its Canvas page, results appear "alongside submissions in SpeedGrader or assignment views", with document scores, sentence highlights and, in its Advanced scan, "video playback of your students' writing". That last part is the closest thing to process tracking in this list, and it's GPTZero's feature, not Instructure's.

Quiz logs in Classic Quizzes

Here's where Canvas does record behaviour on its own. Classic Quizzes can keep an action log with timestamps for "viewed", "answered", "stopped viewing the Canvas quiz-taking page" and "resumed". Instructure's quiz log guide is blunt about its limits, though. Logs are kept for six months, Quiz Log Auditing is a course opt-in feature, and autosave can make some question types look answered at the start. Its own words: "Quiz logs are not intended to validate academic integrity or identify cheating for a quiz."

I like that honesty. I wish more of the detector vendors wrote like that.

The Moderate page in New Quizzes

New Quizzes has a similar log on its Moderate page: start and finish times, each event during the attempt, and "moments of stopping and resuming the quiz". The same warning sits at the top of that guide, saying the logs "should not be used to validate academic integrity or identify occurrences of cheating."

Lockdown browsers and proctoring add-ons

For exams, many schools pair Canvas with Respondus LockDown Browser. Respondus says on its product page that it shows assessments full-screen, blocks other apps, and stops copying and pasting to or from the test; licences also include 200 free seats of Respondus Monitor, which adds webcam proctoring. Again, that's a third-party tool your school has to license. It's not something Canvas does by default.

Things Canvas can't see

It's easy to overestimate an LMS. Based on the Instructure guides above, Canvas on its own doesn't:

• score an essay or discussion post for AI-generated text
• know whether you had ChatGPT, Claude or Gemini open in another window while writing an assignment
• treat a quiz "stopped viewing" event as proof of anything, by Instructure's own rules
• keep Classic Quiz logs beyond six months

What it does have is the context instructors already use: your submission time, your earlier work in the course and any add-on reports. Honestly, I think the comparison with your earlier writing probably catches more people than any percentage, but I can't point to a study that measures that, so treat it as my guess.

Pricing: who pays for detection

Students don't pay for any of this. Schools do, and the detector is a separate licence from Canvas. None of the pages linked here lists a per-school price (checked October 2026).

Performance and independent findings on the detectors schools add

Because Canvas doesn't detect anything itself, the accuracy question is really about Turnitin and friends. And it's contested.

Turnitin has said since 2023 that it aims for "a less than 1% false positive rate", per its blog post on false positives, while admitting that rate "is not zero". Its AI Writing Report guide goes further: the model "may not always be accurate" and "should not be used as the sole basis for adverse actions against a student." Scores from 1 to 19% now show only as an asterisk, because Turnitin found a "higher incidence of false positives" in that range.

Some universities decided that wasn't good enough. Vanderbilt turned the AI detector off in 2023, doing the maths that a 1% false positive rate on its 75,000 papers from 2022 could have meant around 750 wrongly flagged students. The University of Waterloo dropped Turnitin's AI detection in September 2025, saying its own testing found the tool "in more than one instance" flagged human-written text as 100% AI, and that "there are no products on the market that can effectively identify GenAI." That's a strong line from a university, and I think it's closer to the truth than the vendors' 99.9% figures.

GPTZero, for its part, claims "99.9% accuracy" on its Canvas page. That's a vendor number, not an independent audit. Treat it that way.

The fight has also moved to privacy. In October 2026, ProPakistani reported, citing Times Higher Education, that Southampton, Newcastle and Westminster plan to stop using Turnitin, partly over proposed licence changes about using student work to train AI tools. Turnitin has postponed those changes until at least September 2027, according to the same report.

For a wider look at why these scores wobble, see our explainer on whether AI detectors are accurate and how AI detectors work.

Best suited to

This page is for two groups. Students who want to know what's actually visible to their instructor: check your syllabus and your school's academic integrity page, because only your institution knows which add-ons are switched on. And instructors who assumed Canvas had a detector: it doesn't, and Instructure's own quiz guides tell you to talk to your academic integrity office before acting on a log.

If you've been flagged and think it's wrong, ask which tool produced the score and request the full report. Turnitin's own guidance says a score alone shouldn't decide the outcome. That's a fair thing to point to.

Alternatives schools use alongside Canvas

Turnitin (Feedback Studio and Clarity)

Still the most common add-on, as far as I can tell. In May 2026 Turnitin added custom AI settings to Turnitin Clarity, a tool that captures "the entire writing process" and lets instructors set what AI help is allowed per assignment. We cover the detector in detail in our Turnitin AI review.

GPTZero

The most visible Canvas-specific pitch, with SpeedGrader results and writing replay. Our GPTZero review covers its claims and limits.

Copyleaks

Plugs into the same Plagiarism Review slot as Turnitin, with AI and plagiarism checks in one report. If your school is comparing options, our guide to which AI detector is closest to Turnitin sets them side by side.

Verdict

Canvas doesn't detect AI, and I'd rather people stopped saying it does. What it has are logs that Instructure itself says shouldn't be used to prove cheating, plus a slot where schools plug in Turnitin, GPTZero or Copyleaks. If you're a student, the useful question isn't "does Canvas check for AI?" but "which add-on did my school switch on, and what's its policy on false positives?" If you're an instructor, I'd treat any detector score as a reason to talk to a student, never as a verdict. That's what Turnitin's own guide says, and it's probably the only safe reading of these numbers in 2026.

FAQ

What AI detector does Canvas use?

None by default. Canvas has no built-in AI detector. Schools that want one add a third-party tool, most often Turnitin, or GPTZero or Copyleaks, through the Canvas Plagiarism Framework or an LTI app. Results then appear in SpeedGrader or in the vendor's own grading view. Your syllabus or IT help pages should say which one your school uses.

Does Canvas have an AI detector?

No. An Instructure community moderator confirmed that Canvas doesn't give instructors an AI-generated paper detection feature, and GPTZero's Canvas page says the same. Instructure's IgniteAI tools generate rubrics, summarise discussions and translate messages; they don't score writing for AI. Any AI percentage you see comes from an add-on your school licenses.

Does Canvas detect AI?

Not by itself. Canvas records quiz activity, such as when you stopped viewing the quiz page, but Instructure says those logs shouldn't be used to identify cheating, and Classic Quiz logs are kept for only six months. AI detection only happens if your school connects a tool like Turnitin, GPTZero or Copyleaks, and even then accuracy is disputed.

Does Canvas check for AI?

Canvas doesn't check submissions for AI on its own. Whether your work gets checked depends on your institution and each assignment's settings. Turnitin can be present with its AI indicator switched off, as at the University of Minnesota. If you're unsure, ask your instructor which tool, if any, runs on the assignment and how its scores are used.