Flagged as AI but you wrote it? What to do next

Updated October 31, 2026 · 4 min read

Short answer

Don't panic, and don't admit to something you didn't do. An AI detector score is a probability, not proof. Turnitin itself says its AI score "should not be used as the sole basis for adverse actions against a student," and universities such as Melbourne and Syracuse don't accept a detector report on its own as evidence.

Ask to see the report, gather proof of how you wrote the work (version history, drafts, notes, sources), and be ready to explain your argument in your own words. Students have been cleared after false flags, one of them by a court.

Can an AI detector prove you used AI?

No. Detectors estimate how likely it is that text was written by AI. Even the companies that make them say a score is not proof:

Turnitin says its AI score "should not be used as the sole basis for adverse actions against a student" (guide last edited October 2026).
Turnitin states a document-level false positive rate under 1% for documents with over 20% AI writing. In its own words, that means it "might flag a human-written document as AI-written for one out of every 100 fully-human written documents."
At that rate, Vanderbilt University calculated that about 750 of the 75,000 papers it submitted in 2022 could have been wrongly labeled. That was one reason it switched Turnitin's AI detector off in August 2023.
The University of Melbourne tells students that "an AI writing detection report alone is not sufficient evidence for an allegation."
Syracuse University's integrity office "cannot accept cases that only provide AI-detector screenshots, results, or ChatGPT outputs as evidence" and asks for at least three independent points of documentation.

Why does human writing get flagged as AI?

Detectors look for patterns that are typical of AI text. Some kinds of human writing share those patterns:

Plain, predictable wording. In a 2023 Stanford study, seven detectors labeled 61.3% of essays by non-native English speakers as AI on average, while classifying US students' essays accurately. More on that bias.
Text polished with AI tools. In a 2025 study, GPTZero classed 97 to 100% of original research letters as human, but flagged 75 to 85% of them after ChatGPT polished them for readability. A 2026 study found lightly AI-edited abstracts were flagged 38 to 80% of the time.
Repetitive or formulaic text. Turnitin says false positives can include "content without a lot of structural variation, text that literally repeats itself, or text that has been paraphrased without developing new ideas."
Short texts. Detectors are less reliable on short passages. Turnitin doesn't score English documents with under 300 words of qualifying prose.

What to do, step by step

Stay calm and don't sign or admit anything yet. Ask what the concern is, which tool flagged your work and what it reported. At Melbourne, being asked to discuss your work "is not an allegation of academic misconduct."
Ask for the report. Students usually can't see Turnitin's AI report themselves, but instructors can share a PDF copy. Check which passages were highlighted and whether the score was above Turnitin's 20% threshold (scores of 1 to 19% are shown only as an asterisk).
Collect your writing-process evidence. Version history, drafts, notes and sources show how the text came together. See the table below.
Prepare to explain your work. You may be asked "how you developed the argument, what sources you used, how you reached the conclusion you did" (University of Melbourne). Re-read your sources and notes so you can talk through your choices.
Read the policy and deadlines. Find your school's academic integrity policy and the appeal deadline. Cornell students, for example, have ten days to appeal. A student union, advisor or ombuds office can help you prepare.
Add context, not just a second score. A different detector rating your text as human shows that tools disagree. It supports your case but doesn't replace your drafts and version history.

What evidence helps most?

Evidence	Where to find it	Why it helps
Version history	Google Docs: File → Version history. Word: only for files saved in OneDrive or SharePoint.	Timestamps show the text growing over hours or days. A UC Davis student was cleared in 2023 this way.
Drafts and outlines	Saved files, cloud folders, emails to yourself	Show how your argument changed over time.
Notes and sources	Reading notes, annotated PDFs, library history	Show the research behind the text.
Your earlier work	Past assignments and graded papers	Cornell lists "copies of other papers that show evidence of your writing style."
AI chats, if AI use was allowed	Chat history or screenshots	Show exactly what the AI did and didn't write. Cornell suggests saving them.

Writing offline in Word? Microsoft's version history only works for files stored in OneDrive or SharePoint, so save dated copies of your drafts instead.

Have students been cleared after a false flag?

UC Davis, 2023: a student flagged by Turnitin was cleared after sharing step-by-step instructions for reviewing her paper's Google Docs history. The time stamps showed she had written it herself (Rolling Stone).
Adelphi University, 2026: a New York state judge reversed the university's AI-plagiarism finding against a freshman, which had relied on Turnitin, and ordered his record expunged (CBS New York, February 2026).
Australian Catholic University, 2025: its deputy vice-chancellor said every case where Turnitin's AI detection was the only evidence was dismissed immediately. One nursing student still waited six months to be cleared (ABC News).

How can you protect yourself next time?

Write in Google Docs, or in Word with AutoSave on and the file in OneDrive, so version history builds up automatically.
Save numbered drafts and keep your notes.
If your course allows AI, follow its rules for disclosing how you used it and keep the chats.
Check your final draft with an AI detector before you submit, to see which passages read as AI. Here's how to do that.
Don't run your essay through an "AI humanizer." Turnitin says it now looks for text modified by these tools, and humanized text can still be caught.

Questions people also ask

Can a teacher fail me based only on an AI detector?

Policies differ by school, but the tools' makers and many universities say it shouldn't happen. Turnitin says its score should not be the sole basis for adverse action, the University of Melbourne says a detection report alone is not sufficient evidence for an allegation, and Syracuse University's integrity office won't accept a case that only includes detector output. Ask for your school's academic integrity policy in writing.

Does using Grammarly or another writing tool make my essay look AI-written?

Light spelling and grammar fixes usually don't. Features that rewrite whole sentences can: in a 2025 study, 75 to 85% of human-written research letters were flagged as AI after ChatGPT polished them, and a 2026 study found lightly AI-edited abstracts were flagged 38 to 80% of the time. If you used a rewriting feature, say so and keep the original draft.

Should I show my teacher my Google Docs version history?

It's usually the strongest evidence you have, because the timestamps show the text being written over time. Google Docs only lets people with edit access browse earlier versions, so the easiest way is to open it on your own device during a meeting. At Syracuse University, students are not required to share it.

Will checking my essay with another AI detector help my case?

It can show that detectors disagree, which supports the point that one score isn't proof. It doesn't prove you wrote the text, so present it as context next to your drafts and version history, not instead of them.

Sources

By the AI Checker Plus team. We research and draft these guides with help from AI tools, then check every claim against the sources listed above.