Can teachers detect AI?

What AI detector do colleges use?

Turnitin is the detector most colleges already have. It runs inside Canvas, Moodle and Blackboard when the school licenses it. Turnitin says more than 16,000 customers in 185 countries use its services, in its 27 August 2025 press release. Not every customer has AI detection, though. Turnitin's AI writing detection FAQ says the AI indicator is "only available to customers that license Turnitin Originality," or iThenticate 2.0 with the AI add-on, and that administrators can switch it off.

Some schools use other detectors. GPTZero sells a Canvas integration as part of its institutional and enterprise plans; my GPTZero review covers its free tier and paid plans. Copyleaks lists LMS integrations for Canvas, D2L Brightspace, Google Classroom, Moodle, Blackboard, Schoology and Sakai; my Copyleaks review covers its education plans. My AI detector rankings and the GPTZero vs Turnitin comparison set out how they differ on access and limits.

Some schools have turned detection off. Vanderbilt University disabled Turnitin's AI detector in August 2023, saying "we do not believe that AI detection software is an effective tool that should be used." Curtin University in Australia disabled it from 1 January 2026 across all campuses, keeping text matching on and pointing students to Grammarly's Authorship feature instead. MIT Sloan's teaching team publishes a guide titled AI Detectors Don't Work. Here's What to Do Instead.

So the answer depends on your school. Your syllabus, learning platform or academic integrity office can tell you which tools are on. My Turnitin AI detection review explains Turnitin's report in detail, and does Turnitin check for AI covers which models and tools it says it detects.

My take: if your syllabus says nothing about AI detection, ask. Knowing which tools are on, and what the AI rules are, is the single most useful thing a student can find out at the start of a course.

How do teachers know if you use AI?

Teachers rarely rely on one signal. The sources show five common ones.

1. A detector score. Where a school has Turnitin AI detection, every submission gets a percentage that only staff can see. Turnitin says the score "should not be used as the sole basis for action."
2. A mismatch with your earlier work. Teachers compare a submission with your in-class writing, past assignments and the way you write in discussion posts.
3. The writing process. Version history in Google Docs or Word, Turnitin Clarity's writing playback, or a Grammarly Authorship report can show whether text was typed over time or pasted in at once.
4. Content clues. Citations that do not exist, sources that say something different, generic examples that ignore what was taught in class, or facts that are out of date.
5. A conversation. Asking you to explain your argument, your sources and how you reached your conclusion. The University of Melbourne describes exactly this step and says it "is not an allegation of academic misconduct."

Guidance for teachers increasingly favors the last three. Vanderbilt suggested clear AI policies, in-class writing and assignments tied to topics discussed in class rather than detection. MIT Sloan recommends clear policies, open dialogue and assignment design over detector software.

Here is how much weight I think each signal deserves on its own:

Example. Content clues are the signal I find most telling, so here is an illustration I made up. A paragraph cites "Smith and Lee (2021), Journal of Adolescent Wellbeing, 14(3)" for a statistic. The teacher searches and the journal issue has no such article. That is not proof of AI use, since people mis-cite too, but an invented source with a plausible title, volume and issue is a classic pattern in unedited chatbot output, and it is the kind of thing that starts a conversation.

Can Canvas detect AI?

No, not by itself. Canvas is a learning platform; any AI or plagiarism check comes from a tool connected to it.

• What Canvas provides. Instructure's account settings guide describes a Similarity Detection Platform section "used for plagiarism LTI tool integrations," where schools can add a pledge students must accept. It does not describe any built-in AI detector.
• Turnitin in Canvas. Turnitin's FAQ says users can see the AI indicator and report through the LMS they use, but "there is no AI writing indicator or score embedded directly in the LMS user interface." Instructors open the Turnitin report to see it.
• GPTZero in Canvas. GPTZero's Canvas integration page states that "Canvas does not currently include AI detection" and that its integration shows scores alongside submissions in SpeedGrader. It is part of GPTZero's institutional and enterprise plans.
• Grammarly Authorship. Grammarly says students can turn on Authorship in Canvas submissions, Google Docs and Word to record how they wrote. That is a student-controlled record, not a detector.

If you want to know whether your Canvas courses run an AI check, look for Turnitin or GPTZero in the assignment settings or ask your instructor. Canvas quiz logs and page views are activity data, not AI detection.

My take: "Canvas detected my AI" almost always means a connected tool did. Find out which one, because the answer changes what the score means and who can see it.

Can Google Docs detect AI?

No. Google Docs has no AI detector. What it has is version history, and that matters more to many teachers.

Google's help page on finding what has changed in a file explains that you can open File, then Version history, choose a previous version and "find who updated the file and their changes." You can also name up to 40 versions per document. Anyone with edit access to a shared document, including a teacher, can open it.

What version history shows:

• When the text was written. Edits spread across several sessions look different from one large block that appears at once.
• How it grew. Outlines, rewritten paragraphs, deleted sections and moved sentences are visible.
• Who edited it. Each change is linked to an account.

What it does not show is whether the words came from an AI tool. A large paste could be text you wrote in another app, notes from class or copied text. That is why version history is evidence about your process, to be read alongside everything else.

Writing tools add more detail. Turnitin Clarity, a paid add-on to Turnitin Feedback Studio, gives students a composition space and gives instructors "writing process playback" with "flagged points of interest." Grammarly's Authorship records whether text was typed, pasted or AI-assisted, and Grammarly says it does not send the report to anyone unless the student shares it.

My take: version history protects honest students far more often than it catches dishonest ones. I'd write every graded assignment in a tool that keeps it.

Can professors tell if you use ChatGPT?

By reading alone, not reliably. The best evidence comes from a controlled experiment.

Fleckenstein and colleagues ran two studies, published in Computers and Education: Artificial Intelligence in June 2024, in which novice and experienced teachers could not identify texts generated by ChatGPT among student essays. More experienced teachers made "more differentiated and more accurate judgments," but both groups were overconfident. The experiment covered 89 novice and 200 experienced teachers. The authors also found that AI-generated essays tended to be rated more positively than student texts.

Detectors do not close the gap reliably either:

• They miss AI text. Turnitin says a 50% score could contain as much as 65% AI writing, because it tunes for few false positives. Perkins and colleagues found seven detectors had a mean accuracy of 39.5% on unedited AI text in 2023 tests.
• They can flag human text. Liang and colleagues found seven detectors flagged non-native English essays as AI at an average rate of 61.3%.
• They disagree. The same text can score very differently across tools.

So when professors do conclude that ChatGPT was used, it is usually because several things line up: a detector score, a sudden change in style, invented citations, an empty version history and a student who cannot explain the work. That is also why the reverse matters. If you wrote the work yourself, the same kinds of evidence can show it. My guide to AI detector false positives covers why honest work gets flagged, and my explainer on how AI detectors work shows why plain writing can trip a detector.

Using a rewriting tool to make AI text read "more human" does not change this picture. Teachers are weighing process and understanding, not just style, and schools set their own rules on these tools. My guide on whether using an AI humanizer is cheating sets out what real policies say.

Keep your drafts to show your process

The strongest thing an honest student can do is keep a record of how the work was made. Universities name the same evidence again and again.

• Write in a tool with history. Google Docs and Word online keep version history automatically. Name versions at key points, such as "outline," "first draft" and "final."
• Keep your notes and sources. Reading notes, highlighted articles, an outline and a list of sources show the thinking behind the essay.
• Save the feedback you used. Comments from a tutor, writing center or classmate show how the draft changed.
• Turn on a process tool if your course offers one. Turnitin Clarity or Grammarly Authorship records the writing as it happens.
• Know your course's AI rules. Schools set their own policies. If AI use is allowed, disclose it the way your course asks; Vanderbilt points instructors to APA and MLA guidance on citing AI.
• Be ready to explain the work. Melbourne says students may be asked how they developed the argument, which sources they used and how they reached the conclusion.

If you are flagged, ask for the report and its date, bring your drafts and version history, and ask your student union or advocacy service for help with the process. My guide to whether Turnitin can detect humanized AI text explains why rewriting tools are a poor answer to a false flag. Educators building a classroom policy can start with my AI tools page for educators. More detection guides are in the HumanRank guides hub.

How we checked this

• Read on 29 October 2026: Turnitin's AI writing detection FAQ, its Turnitin Clarity product page and its 27 August 2025 press release; GPTZero's Canvas integration page; Instructure's account settings guide; Google's Docs Editors Help page on version history; Grammarly's Authorship page.
• Studies read: Fleckenstein et al. (Computers and Education: Artificial Intelligence, 2024), Liang et al. (Patterns, 2023) and Perkins et al. (International Journal of Educational Technology in Higher Education, 2024).
• University guidance: Vanderbilt University (August 2023), Curtin University (September 2025 notice), the University of Melbourne's student advice and MIT Sloan Teaching and Learning Technologies.
• Method: I attributed vendor statements as claims and kept my own judgement in the "My take" lines. The site's general approach is on the HumanRank ranking methodology page.

Frequently asked questions

Can teachers tell if you use AI? Not reliably by reading. A 2024 study found novice and experienced teachers could not pick out ChatGPT essays and were overconfident. Teachers usually combine a detector score, earlier work, version history and a conversation.

What AI detector do most universities use? Turnitin, where the university licenses Turnitin Originality and has AI detection switched on. Some use GPTZero or Copyleaks, and some, including Vanderbilt and Curtin, have disabled AI detection.

Can Canvas detect ChatGPT? Canvas has no built-in AI detector. It shows results from connected tools, such as Turnitin's report or GPTZero's SpeedGrader integration, if your school has set them up.

Can teachers see your Google Docs history? Yes, if they have edit access to the document. Version history shows who changed the file, when and what changed.

Can teachers see if you copy and paste? Version history can show a large block of text appearing at once, and tools such as Turnitin Clarity and Grammarly Authorship label pasted text. They do not show where the pasted text came from.

Do teachers use AI detectors on every assignment? Where Turnitin AI detection is enabled, it runs automatically on every Turnitin submission. Whether a teacher acts on it is their decision under the school's policy.

What should I do if a teacher thinks I used AI? Stay calm, ask for the report and its date, and bring drafts, version history, notes and sources. Being asked to explain your work is not, on its own, an allegation of misconduct.

REVIEW NOTEI checked this guide on 29 October 2026 against Turnitin, GPTZero, Instructure, Google and Grammarly documentation, three published studies and guidance from four universities. Every number on this page belongs to the named source, and the opinions are mine.