# Falsely Accused of Using AI? Here’s What to Do in the First 48 Hours

> A calm, practical guide for students who wrote their own work and got flagged anyway — from gathering evidence to asking the right questions.

- Author: [Priya Raman](https://www.detectionguide.com/authors/priya-raman/), Education correspondent
- Published: 2026-09-30
- Section: Classroom
- URL: https://www.detectionguide.com/articles/falsely-accused-of-using-ai/

## Key takeaways

- Being flagged is not the same as being found responsible, and a detector score on its own is weak evidence.
- Don’t edit or “humanize” the work you submitted. Leave it exactly as it is and show how it was made.
- Your strongest evidence is the trail your writing left: version history in Google Docs or Word on OneDrive, drafts, notes, research history and messages.
- Ask what evidence besides the score is being used, and check what your institution’s policy says about detector results.

The email usually arrives at a bad time. Your essay, the one you spent a week on, has been flagged as likely AI-generated. There's a percentage. There may be a deadline to respond.

First: breathe. Being flagged is not the same as being found responsible, and a detector score on its own is weak evidence. Here's how to respond in a way that is calm, organized and hard to dismiss.

## Don't panic, and don't "fix" the essay

The worst move is the instinctive one: running your essay through a humanizer or rewriting it so it "passes." Any edit after the fact muddies the record, and if it comes out later, it looks far worse than the original flag.

Leave the submitted version exactly as it is. Your job now is to show how it was made.

## Gather your evidence

Writing leaves a trail. Most of it is already sitting in your accounts.

- **Version history.** Google Docs (File → Version history) and Microsoft Word files saved to OneDrive keep timestamped revisions. A document that grew over days, with deletions and rewrites, tells a story no chatbot transcript can.
- **Drafts and outlines.** Earlier files, handwritten notes, photos of a whiteboard, an outline in your notes app.
- **Research trail.** Library database history, PDFs you annotated, browser history around the time you were writing.
- **Messages.** Emails to your instructor or TA, group chats where you talked through ideas, feedback from a writing center.
- **Previous work.** Earlier essays in the same course show your voice and style.

Export or screenshot what you can and put it in one folder with dates.

> **The 48-hour checklist**
>
> - Save the original email and note any response deadline.
> - Export version history and collect drafts, notes and sources.
> - Find your institution’s academic integrity policy and read the process section.
> - Reply briefly and politely: you wrote the work and would welcome a meeting.
> - Ask whether you can bring a support person or student advocate.

## Ask the right questions

When you respond, you're allowed to ask how the decision is being made. Polite, specific questions are your friends:

1. What evidence, besides the detector score, is being considered?
2. Which tool produced the score, and which version or setting?
3. Can I see the highlighted passages and the full report?
4. What does the institution's policy say about using detector results?

That last question matters. Most detector vendors' own guidance says their scores should not be the sole basis for disciplinary action, and many institutions have written the same caution into their policies. You're not being difficult by pointing that out — you're pointing to the rules.

## Offer to talk about your work

The strongest evidence of authorship is often a conversation. Offer to walk through your argument, explain why you chose particular sources, or talk about a paragraph you struggled with. People who wrote something can usually discuss it in depth. That's exactly what a detector can't measure.

## Get support

Many universities have an ombudsperson, student advocacy office or student union adviser who has seen these cases before. If English isn't your first language, mention it — research has found that some detectors flag non-native writers disproportionately, and a fair process should take that into account.

## Afterwards

Whatever the outcome, keep writing in tools that save version history, keep your notes, and save your drafts. It feels absurd to document your own honesty. Right now, it's also the best protection you have.

## Common questions

### What should I do if I’m falsely accused of using AI?

Stay calm and don’t change the submitted work. Save the email and any response deadline, export your version history, collect drafts, notes and sources, read your institution’s academic integrity policy, and reply politely that you wrote the work and would welcome a meeting.

### How can I prove I wrote my essay myself?

Show the trail your writing left: timestamped version history in Google Docs or Word files saved to OneDrive, earlier drafts and outlines, research history, messages about the work, and previous essays in the same course. Offering to talk through your argument and sources is strong evidence too.

### Should an AI detector score alone be used to discipline a student?

Most detector vendors’ own guidance says no: their scores should not be the sole basis for disciplinary action. Many institutions have written the same caution into their policies, so ask what other evidence is being considered.