False positives · a person has accused you; the conversation itself
Someone accused you of using AI. Here is the conversation you're actually in.
There is a conversation happening right now, or about to happen, between you and whoever raised this. This page is about that conversation: what the process actually looks like from the other side, and what to say when someone you need to trust you is looking at you like they don't.
Being accused of using AI is not the same event as being found responsible for it, and treating the accusation like a verdict, panicking and over-explaining, is what turns a flag into a bigger problem than it needs to be. Right now someone has raised a concern and wants to talk to you about it. That is a different thing than a formal charge, and you can ask directly which one you're in instead of guessing. What you say in this first conversation gets read for tone as much as content, and three specific habits, over-apologizing, offering an alibi before being asked, and explaining your writing process in more detail than the question called for, all read as the behavior of someone with something to hide, even when there is nothing to hide.
Why this happens
Instructors who raise an AI concern are usually reacting to a mismatch, not a certainty: the essay reads differently than your in-class writing, or a detector score crossed some internal threshold they've decided to trust. Neither is proof, but both feel like evidence to the person holding them, because they have no easy way to distinguish a student who wrote carefully from a machine that wrote fluently. Statistical detectors measure how predictable your word choices are, and honest writing that follows a taught structure, hedges appropriately, or comes from a non-native speaker who was taught formal register, all compress the natural variation the detector is built to notice. The instructor is not lying about being concerned. They are trusting a tool that is less certain than its output looks.
Where honest writing and machine writing genuinely overlap
These are real patterns, not coincidences. Each one is something a careful human writer in this situation has good reason to do, and something a model does for entirely different reasons.
- Over-explaining the writing process unprompted
- A student who launches into a five-minute account of every draft before being asked reads as rehearsed to someone already suspicious, even when the account is entirely true and the rehearsal is just anxiety.
- Apologizing before being asked to
- 'I'm sorry, I know how this looks' sounds like an admission to a listener primed to hear one, even when it is only nerves talking.
- Matching the accusation's intensity instead of its content
- Getting defensive or upset in response to a calm, procedural question reads as guilt to an instructor who was asking a routine process question, not making a final judgment.
- Volunteering irrelevant detail as proof of honesty
- Naming exactly which website, which hour, which coffee shop, in an effort to sound credible, often reads as over-rehearsed rather than as reassuring, because genuine memory of routine work is rarely that precise.
What actually helps
Ask, specifically and calmly, what raised the concern: a detector score, a style mismatch, something else. Do not answer a vague accusation with a long explanation; answer a specific one with a specific response. If drafts, notes, or version history exist, offer them once, plainly, rather than repeating the offer. Bring a second document if your school allows it, an advisor, an ombudsperson, a parent for younger students. None of this proves innocence by itself. It shows you are willing to have the conversation the process actually requires, which is most of what a first-stage meeting is checking for.
What the flattening looks like
On the left, honest writing with the specifics drained out of it, which is the shape that reads as machine-written. On the right, the same content with the writer's own detail restored. The voice is The Steady Hand
I understand your concern and I want to be completely transparent with you. I wrote every single word of this essay myself, without exception, and I would never do something like use AI, that's just not who I am as a person, and I've always taken academic honesty very seriously throughout my entire time as a student here.
I hear you, and I want to be clear. I wrote every word of this essay myself. I would not use AI for something like that — it's not who I am. I've always taken academic honesty seriously, the whole time I've been a student here.
Steps worth taking
- Ask what triggered the concern A detector score and a style mismatch call for different responses. Find out which one you're dealing with before you say anything else.
- Answer the specific question, not the general one If they ask about a paragraph, answer about the paragraph. Resist expanding a narrow question into a defense of your whole record.
- Offer evidence once Mention your drafts or version history a single time, plainly. Repeating the offer reads as pleading, not as confidence.
- Know your school's actual process before the meeting Most academic-integrity policies are published. Read yours so you know whether this conversation is informal or already on the record.
Questions
What to do if falsely accused of using AI?
Find out specifically what raised the concern, gather your drafts and any version history, and answer the specific question you're asked rather than delivering a general defense of your character.
How to respond to an AI accusation?
Calmly, specifically, and once. State plainly that you wrote it, ask what was flagged, and offer your process evidence a single time rather than repeating it.
Can I get in serious trouble for something I didn't do?
It depends on your school's process and how the conversation goes. Ask directly whether you're in an informal conversation or a formal charge, since that changes what's at stake, and staying calm and specific matters more in the first than in the second.
What do I say if I don't have any drafts saved?
Say that plainly too. 'I don't have earlier drafts saved, I wrote this in one sitting' is a normal answer many honest writers give. Trying to reconstruct evidence you don't have is worse than admitting you don't have it.
Why does my professor think my essay is AI when other students write similarly and weren't flagged?
Detector scores are inconsistent between essays for reasons that have nothing to do with quality, including length, topic, and how closely the essay follows a taught structure. Inconsistent flagging across similar essays is itself worth raising, calmly, as part of the conversation.
underslop edits drafts for voice. It does not check, score, or contest anything, and nothing on this page will change a result you have already been given. What it can do is show you which patterns in a piece of writing read as machine-written, using the same open lint the rest of the site runs on.