Falsely Accused of Using AI? What to Do Next

plagiarism-checker-online.net Editorial Team  |  Updated October 3, 2026

An email arrives: your professor believes your paper was written by AI. You wrote it yourself, and now you have a meeting, maybe a hearing, and a stomach full of dread. This guide is for that moment. It covers what to do in the first 24 hours, why detectors flag human writing, what evidence helps and where to find it, how to prepare for the meeting, and how appeals work. It doesn't promise an outcome, because no article can. It does give you steps that make a calm, factual response easier.

This is general information, not legal advice. Rules differ by school, so your own policy always comes first.

The First 24 Hours: Steps 1 to 3

1. Step 1: Don't edit, delete or "fix" anything. Leave the submitted paper and its drafts as they are. Work on copies if you need to open them. Don't run the paper through a rewriting or "humanizer" tool: it changes the text under review and can look like an attempt to hide something. Don't send a heated email either. A short, polite message asking for details is enough.
2. Step 2: Save everything. Export or screenshot the document's version history before anything changes. In Google Docs it's under File, Version history. In Word, documents saved to OneDrive or SharePoint have Version History under File, Info. Copy your outline, notes, source PDFs and any messages about the paper into one folder, and note the date you did it.
3. Step 3: Get the accusation in writing. Ask which passages are flagged, which tool or observation the concern is based on, what score or date it refers to, which policy it falls under, and when and where the meeting will be. Some schools require this by policy: San Jose State's academic integrity policy, as summarized by its library, requires a faculty member who suspects a violation to explain the allegation and the supporting evidence and give the student a chance to respond.

A message you can adapt: "Thank you for letting me know. I wrote this paper myself and would like to explain how I worked. Could you send me the flagged passages, the tool or report you used, and the policy section this falls under? I'll bring my drafts and notes to the meeting."

Why False Positives Happen

AI detectors estimate how predictable your text is and how it compares with patterns typical of machine writing. Predictable, tidy prose can trigger them, and that includes plenty of human writing. The best-known study is Liang et al. (2023) in the peer-reviewed journal Patterns. The authors ran seven popular detectors on 91 TOEFL essays written by non-native English speakers and on 88 US eighth-grade essays. The detectors classified the US essays accurately but labeled more than half of the TOEFL essays as AI-generated, with an average false positive rate of 61.3%. All seven detectors agreed that 19.8% of the human-written TOEFL essays were AI-written, and at least one detector flagged 97.8% of them. The authors link this to text perplexity: writers with a more limited range of expression produce more predictable text, which detectors read as machine-like.

Know the study's limits before you cite it in a meeting. The sample is small, the essays come from one source, and the detectors were the 2023 versions, which have been updated since. It doesn't show that every detector flags every non-native writer today. It shows that a detector score can be wrong for human writing, which is the point you need. Turnitin itself withholds the percentage for results in the 1% to 19% range because of false positive risk, shows an asterisk instead, and states that an AI writing score should not be used as the sole basis for adverse action against a student. For more on the research, see AI detector reliability in 2026, AI detector bias and international students and ChatGPT detection accuracy.

Step 4: Read the Policy and the Procedure

Two documents matter: the rule you're accused of breaking and the process for handling the accusation.

• The rule. Find the AI statement in your syllabus or assignment sheet and the honor code or academic integrity policy. Is AI use banned, allowed with citation, or allowed for some tasks? Our overview of university AI policies explains the common approaches.
• The process. Look for who decides, what the stages are, and what the deadlines are. Michigan State's ombudsperson page, for instance, lays out a ladder: understand the accusation, meet the instructor, speak with the department chair or director, and if needed request a hearing before the University Academic Integrity Hearing Board. The same page says students have 10 class days to file a written request for an academic grievance hearing in the cases it describes. Other schools set different windows, and missing one can cost you a stage of review.
• The standard. State what level of proof your school requires only if the policy publishes it. If it does, quote it. If it doesn't, don't guess.
• The detector rules. Some institutions limit how detector results can be used. San Jose State's policy says academic sanctions may not be imposed based solely on AI-detection software results. Washington State's provost wrote in a February 2026 memo that the university does not allow any AI detector as the sole source of support for a case against a student and cancelled its Turnitin AI detection contract. The University of Pittsburgh's teaching center says detectors are not accurate enough to prove that a student violated academic integrity policies. If your school has a similar rule, cite it politely.

Check that every source in your paper exists and says what you cite it for. The Pittsburgh teaching center notes that hallucinated sources can themselves be an integrity violation. If one of your citations came from a chatbot suggestion and you never verified it, find out now. Our Source Check compares references against Crossref.

Step 5: Gather Your Evidence

The goal is to show how the paper came to exist. The University of Maryland, Baltimore County's Division of Information Technology describes the logic behind version history: instead of trying to detect AI, look for the absence of originality. Writing usually grows in many small edits over time, while pasted material tends to show up in large blocks all at once. UMBC also notes that some work happens offline, such as handwritten notes or work in other editors, so version history alone doesn't always tell the whole story. That's why you collect several kinds of evidence.

Never fabricate, backdate or "recreate" evidence. Don't type up a fake version history or change file dates. It can turn a false accusation into a real violation, and it's the fastest way to lose a case you could have won with honest material. If you don't have much, say so, and rely on what you do have plus your ability to talk through the paper.

Timing matters too. At the University of Maryland, an appeal based on new evidence succeeds only if the evidence could not reasonably have been found before the original proceeding. Bring what you have to the first meeting, not after a decision.

Step 6: Prepare for the Meeting

Treat it as a conversation about your paper, not a trial. Print your evidence, bring notes, and expect to talk about your argument and sources. San Jose State's guidance for faculty suggests talking with the student about their work and writing process and comparing the paper with their other work, so be ready to do exactly that.

Talking-point template (adapt it):

1. Open: "I wrote this paper myself. I'd like to show you how I worked and answer any questions."
2. Walk through your process: how you picked the topic, which sources you read first, when you outlined, how many drafts you wrote and when.
3. Show the evidence: version history, notes, annotated sources. Hand over copies, keep the originals.
4. Offer to discuss the content: invite questions about the argument, why you chose certain sources, and what you would change.
5. Offer a short supervised writing sample on a related prompt, if the instructor wants one.
6. Ask: which passages were flagged, which tool and report were used, what happens next, and by when.
7. Close in writing: within a day, email a short summary of what was said and any next steps, and keep it factual.

Ask whether you may bring a support person or an advisor. Rules about that differ between schools. Keep your tone neutral even if the other side doesn't. A calm record helps you at every later stage.

Step 7: Appeal and Escalate

If the conversation doesn't resolve it, use the next stage of your school's process. A typical order, as the Michigan State ombudsperson page lays it out, is instructor, then department chair or director, then a hearing board. The University Ombudsperson at Michigan State describes itself as a confidential resource that doesn't act as an advocate for students but advocates for fairness and due process. Many universities have an ombuds office or a student conduct office that can explain the process, so ask.

Two things to know about formal appeals. They often run on strict deadlines: the University of Maryland's student conduct office requires an appeal within five days of the notice of the final determination, with limited extensions. And they usually aren't a second review of the facts. Maryland's code says appeals are not intended to allow a second review of whether the student is responsible, and lists narrower grounds such as substantial procedural error, disproportionate sanction, an arbitrary and capricious decision, and new evidence. Build your case at the first stage and keep your tone factual at each one. Describe what happened and what you're asking for, and leave out accusations about the instructor's motives.

Step 8: Protect Yourself Next Time

• Write in a tool with version history (Google Docs, or Word on OneDrive) from the first draft, not as a paste of a finished text.
• Keep your process files: notes, outlines, annotated PDFs, dated drafts.
• Follow the AI policy to the letter. If AI use is allowed, keep a log of what you used it for and write a disclosure. Our free AI usage statement tool turns your notes into a statement and a usage log.
• Revise in your own voice. Feedback tools that don't write for you can help. Our Writing Coach is a rule-based style checker for long sentences, passive voice and filler words. It's not a detector and can't predict one.
• Run your own check before you submit. An AI scan shows how a detector reads your text, and a plagiarism scan shows any overlap with published sources. Keep the dated reports with your files. They are signals, not proof, and different detectors can disagree.

For a bigger picture of what's at stake in integrity cases, see plagiarism consequences.

If assistance is permitted, document it accurately while you work. Our AI disclosure statement templates explain how to record outlining, editing, code assistance or generated text without overstating your own checks. A statement supports transparency; it doesn't guarantee that a concern will be dismissed.

If You Did Use AI in an Allowed Way

Sometimes the accusation lands on a paper where you did use AI, within the rules, and the question is whether it was disclosed. Say so plainly. Show your prompts, what you kept or rewrote, and the policy section that allows it. Citing or disclosing correctly is covered in how to cite ChatGPT in APA, MLA and Chicago, and the AI usage statement tool helps with the disclosure. If you used AI in a way the rules don't allow, don't deny it. Statements that your own files contradict make every other point you make harder to believe, and many integrity processes have routes for students who are straightforward about what happened.

Frequently Asked Questions

Can a professor fail me based only on an AI detector score?

Many institutions say they shouldn't. San Jose State's policy states that academic sanctions may not be imposed based solely on AI-detection results, Washington State's provost memo bars using any AI detector as the sole source of support for a case, and the University of Pittsburgh's teaching center says detectors are not accurate enough to prove a violation. Turnitin itself says its AI score should not be the sole basis for adverse action. Policies differ by school, though, and other evidence such as fabricated citations can still count, so read your own.

What proof shows I wrote my essay myself?

Evidence of your process: version history from Google Docs or Word, an outline and notes, annotated source PDFs, dated drafts, messages with a tutor or teaching assistant, and earlier writing samples. Being able to explain your argument and sources out loud is also strong. None of it is a guarantee, and you should never fabricate or backdate anything.

What is a false positive AI rate?

It is the share of human-written texts that a detector wrongly labels as AI-generated. Rates vary by tool, text and writer. Turnitin reports a document-level false positive rate below 1 percent for documents with 20 percent or more AI writing, which is the vendor's own figure. A 2023 peer-reviewed study found that seven detectors flagged non-native English essays at an average false positive rate of 61.3 percent.

Should I hire a lawyer for an AI accusation?

Most cases are handled inside the school's own process, and the university ombuds office and student conduct office can explain it. If you face a serious sanction such as suspension or expulsion, it is reasonable to ask about student legal services or consult a lawyer. This is general information, not legal advice.

Can I run an AI check on my own paper first?

Yes. An AI scan shows you how a detector reads your text before an instructor's tool does, and a dated report adds to your record. Treat the result as a signal and not as proof in either direction, because different detectors can disagree on the same text. Our AI Scan costs $0.29 per page with a $0.90 minimum.

Sources

• Liang, W., Yuksekgonul, M., Mao, Y., Wu, E., & Zou, J. (2023). GPT detectors are biased against non-native English writers. Patterns, 4(7), 100779. doi.org/10.1016/j.patter.2023.100779
• Turnitin Guides: AI writing detection model (asterisk for 1% to 19%, score not to be used as sole basis for adverse action) and AI writing detection capabilities FAQs
• UMBC Division of Information Technology: Using Google Doc "Versions" to Detect Student Originality vs. AI Abuse
• San José State University Library: I Don't Allow Generative AI. What If I Suspect Unauthorized Use?
• University of Pittsburgh, University Center for Teaching and Learning: Encouraging Academic Integrity (statement about AI detectors)
• Washington State University, Office of the Provost: Cancellation of Turnitin AI Detection software and the use of AI detectors for academic integrity (memo to instructors, February 11, 2026)
• Michigan State University, Office of the University Ombudsperson: How to Contest an Allegation of Academic Misconduct
• University of Maryland, Office of Student Conduct: Academic Integrity Appeals

This article is part of our Academic Integrity Guide.