How to check if an essay was written by ChatGPT

To check if an essay was written by ChatGPT, compare it with the student's own earlier work, verify its facts and references, run it through an AI detector as one signal, and look at the drafts and version history before talking to the student about how they wrote it. No single step proves anything. The aim is to gather enough evidence for a fair decision, and students can use the same steps to check their own work before handing it in.

Can you ask ChatGPT whether it wrote an essay?

No. It's a common mistake, so it's worth clearing up first. OpenAI says ChatGPT has no knowledge of what content is AI-generated, and that its replies to questions like "did you write this essay?" are random and have no basis in fact. A chatbot that says "yes, I wrote this" about a student's own work is making it up.

How to check if an essay was written by ChatGPT: steps for teachers

Work through the evidence in order, from the cheapest checks to the most direct ones. Keep notes as you go, because a fair process is one you can explain afterwards.

1. Start with the assignment rules. Was any AI use allowed, such as brainstorming or grammar help? Turnitin's own guidance for instructors says to begin with your institution's policy on generative AI. A student can't break a rule that was never set.
2. Read it next to the student's other work. In-class writing, earlier essays and emails are your best baseline. A sudden jump in fluency is worth noticing, though students do improve, and some get help from tutors or friends.
3. Check facts and references. Look up two or three citations. Sources that don't exist, page numbers that don't match, or quotes nobody said are strong clues and easy to show to the student.
4. Run a detector, and know its limits. Each tool has its own rules: Turnitin's indicator, for example, needs 300 words of prose and hides low scores behind an asterisk (more on how Turnitin works). Our AI detector for essays works from 50 words, is free, needs no account and never uploads the text. Like any detector, it can be wrong in both directions, so treat a result as a reason to look closer, not as proof; our methodology page explains how it is tested.
5. Look at how the essay was written. If the student worked in Google Docs or Word, the version history shows how the document grew. More on this below.
6. Talk to the student. Ask them to explain their argument, a source they used, or why they structured a paragraph the way they did. Turnitin suggests opening with "Explain your process for completing this assignment." A student who did the work can usually do this without much trouble.

What does version history show, and what doesn't it show?

Version history is the closest thing to a record of the writing itself. In Google Docs you open it by clicking "Last edit" at the top right of the document. You need permission to edit the file to browse earlier versions, so ask the student to share it with edit access or make a copy for you. Each version shows who changed the document and what they changed. In Microsoft 365, version history works for files stored in OneDrive or SharePoint: click the file's title and choose "Version history".

What to look for:

• An essay that grew over several sessions, with rewrites, moved paragraphs and deleted false starts, is what most genuine writing looks like.
• An essay that appears almost complete in one or two versions deserves a question. It isn't proof: plenty of students draft in another app, on paper or in a notes file, then paste the result in.
• Missing history isn't proof either. Google merges some revisions automatically, and a file's owner can delete its version history.

Treat version history the way you would treat any other evidence. It can support a student's account strongly, and a gap in it is a reason to ask questions before drawing any conclusion.

How can students check their own essay before handing it in?

If you wrote your essay yourself, you may still worry about being flagged. A few habits protect you far better than any score.

• Run a private check. Paste your essay into a detector that doesn't keep a copy. Ours runs in your browser, so an unpublished essay never leaves your device. Different tools give different scores, so a clean result from one doesn't predict what another will say.
• If a passage is flagged, look at why. Check the sentence map. Formulaic openings, very even sentences and textbook phrasing can look machine-like. Rewrite in your own words if you want to, but don't feed your work through an AI "humanizer". That puts AI-written text into an essay that was yours.
• Be careful with AI editing tools. Grammar fixes are one thing; text drafted or paraphrased by an AI feature is another, and detectors treat it that way. Turnitin's own position on Grammarly is in our Turnitin article.
• Keep your process. Write in Google Docs or Word with version history on, and keep notes, outlines and reading lists. If you are ever questioned, this is your strongest evidence. Our guide to being wrongly flagged by an AI detector covers the rest.
• Disclose permitted use. If your teacher allowed AI for brainstorming or feedback, say what you used it for. A short note avoids most misunderstandings.

How reliable are AI detectors on student essays?

It depends heavily on the tool and on the kind of AI use. In a June 2026 test on long academic papers, three of four detectors, Turnitin among them, gave most of the 40 papers written entirely by ChatGPT an AI score under 20%, while Pangram misclassified only one; our Turnitin article has the breakdown. False positives are the other risk. Plain, careful prose, the kind English learners are taught to write, can look machine-made to a detector; a 2023 Stanford study made that case forcefully, as we explain in AI detectors and non-native English. A student can be flagged for reasons that have nothing to do with ChatGPT.

So use a detector to decide where to look, and let the drafts and the conversation decide the outcome. Turnitin says the same about its own tool: the percentage should not be the sole basis for action. If you teach, our page for teachers explains how to use our report in class.

Sources

1. OpenAI Help Center: How can educators respond to students presenting AI-generated content as their own?, checked October 2026
2. Turnitin Guides: What should I do if the AI writing percentage is high?, checked October 2026
3. Turnitin Guides: What questions do I ask my student if AI is detected?, checked October 2026
4. Turnitin Guides: Turnitin's AI writing detection capabilities FAQs, checked October 2026
5. Turnitin Guides: File requirements for an AI Writing Report, checked October 2026
6. Google Docs Editors Help: Find what's changed in a file, checked October 2026
7. Microsoft Support: View previous versions of Office files, checked October 2026
8. Van Vlasselaer, Van Droogenbroeck and Spruyt, Who wrote this? Evaluating the reliability of AI detection tools in higher education (International Journal for Educational Integrity, 2026), checked October 2026
9. Liang et al., GPT detectors are biased against non-native English writers (Patterns, 2023), checked October 2026