Reported for AI Use: What an Academic Integrity Case Actually Looks Like, Step by Step

If you've been reported for AI use, here's what typically happens: a formal report goes to your institution's integrity office, you're notified in writing, you meet with the instructor or an officer, you get a window to respond, and — if it isn't resolved there — the case goes to a hearing, a finding, and possibly an appeal. At almost every stage, you have more rights than you think, and the single most useful thing you can do is stop panicking and start collecting your drafts.

One thing before anything else: this article describes common patterns across universities, not the rules of your university, and it is not legal advice. Every institution runs its own process under its own code, and the differences matter. Your first practical task — before you reply to any email — is to find your school's academic integrity policy and read the section on procedure. Everything below is a map of the usual terrain, so that document makes sense when you open it.

And a word to the two kinds of readers here. If you didn't use AI and got flagged anyway, you're not paranoid and you're not alone — false positives are a documented, systemic problem, and your case is more defensible than it feels right now. If you did use AI beyond what the course allowed, this guide is still for you: the process exists to hear you out, honesty tends to produce dramatically better outcomes than doubling down, and you're entitled to a fair procedure either way.

Stage 1: The report

Most cases begin with an instructor. They saw a Turnitin AI score, or the essay didn't sound like your in-class writing, or a paragraph cited a source that doesn't exist. At many institutions, instructors are required to report suspected violations to a central office rather than handle them privately; at others, they can resolve minor cases directly and only escalate serious or contested ones.

Two things about this stage are worth knowing even though you're not in the room. First, the instructor usually has to attach evidence — the flagged assignment, the detector report, a written explanation of their concern. Second, at many schools a formal report triggers a records check: if you have a prior finding, the process and the stakes often escalate. A first offense and a second offense are frequently different procedural universes.

You'll typically learn about the report through a formal notice: an email from the instructor asking you to meet, or a letter from an office with a name like Student Conduct, Academic Integrity, or the Dean of Students. Read that notice slowly. It usually states what you're accused of, under which section of the code, and what happens next — and buried in it are your first deadlines.

What to do at this stage: nothing fast. Don't fire back a defensive reply at 1 a.m. Don't delete anything — ever, at any stage; destroying drafts or history looks like consciousness of guilt and can be a separate violation. Start preserving instead: your document version history, your notes, your browser history, your sources.

Stage 2: The initial meeting

Nearly every process starts with a conversation — usually with the instructor, sometimes with an integrity officer. Depending on the institution, this meeting may be informal ("help me understand how you wrote this") or an official first step with paperwork.

Common rights at this stage — again, patterns, not guarantees; check your code:

• You're generally entitled to know what you're accused of and to see the evidence, including the detector report, before or during the meeting.
• You can usually ask to reschedule if the notice gave you little time to prepare. A short, polite request for a few days is normal and rarely refused.
• At many institutions you may bring a support person or advisor — sometimes any member of the university community, sometimes a trained integrity advisor, occasionally an attorney in a non-speaking role. The policy will say.
• You're typically not required to answer on the spot. "I'd like to review the report and respond in writing" is a legitimate sentence.

The meeting matters more than students expect, in both directions. Many cases end here — an instructor genuinely persuaded by your drafts, or a student who acknowledges what happened and accepts an informal resolution. But what you say is generally on the record. If you did nothing wrong, do not apologize vaguely to make the conversation end; a mumbled "I guess I might have used it a little" from a panicked innocent student is the single most self-destructive sentence in this whole process. If you did violate the policy, be careful about the opposite failure — an elaborate false story that unravels usually converts a recoverable situation into a severe one.

A practical script for the innocent: bring your version history, walk through it chronologically, explain your writing process in concrete detail (where you worked, what sources you gathered, what you cut), and offer to discuss the essay's content in depth. Familiarity with your own argument is evidence no detector can generate.

Stage 3: The response window

If the case proceeds formally, you'll normally get a defined period — often somewhere between a few days and two weeks — to submit a written response, accept or contest the allegation, and elect your procedural route (some schools let you choose between an administrative resolution and a full panel hearing).

This window is where cases are won, because it's where you assemble evidence at your own pace. Prioritize in this order:

1. Version history. Google Docs (File → Version history), Word/OneDrive autosave versions, or file timestamps. A document that grew over eleven sessions across six days is powerful; a document pasted in whole is what the accusation predicts. This is the closest thing to dispositive evidence a student can produce.
2. Process artifacts. Outlines, notes, annotated readings, library checkouts, search history, texts to a friend about the assignment, a writing-center appointment.
3. Your writing baseline. Earlier graded work from this course or others, ideally in-class or proctored writing, showing that the flagged paper sounds like you.
4. Context on the evidence against you. If the case rests on a detector score, the score's limitations are legitimately part of your response — see below.

If your school allows an advisor, this is the stage to get one (more on that shortly).

Stage 4: The hearing

Contested cases typically reach a hearing or panel: often a small board of faculty, administrators, and sometimes students. Format varies enormously — some are conversational, some quasi-judicial — but the usual shape is: the case against you is presented, you respond, the panel asks questions, sometimes each side offers a closing.

Typical rights and features at this stage, where the pattern holds:

• Advance notice of the hearing date, the panel's composition, and the evidence packet — usually with a chance to object to a panelist with a conflict of interest.
• The right to be heard: to present your evidence, and often to question (directly or through the chair) the basis of the accusation.
• An advisor present, under whatever speaking rules the code sets.
• A decision made on a defined standard of proof. At most U.S. institutions this is the preponderance of the evidence — "more likely than not" — not the criminal "beyond a reasonable doubt." Know which standard your school uses, because it shapes your whole argument: your job is usually to show your authorship is at least as likely as the alternative.

Panels are human. They respond to organization, calm, and specifics. A student who arrives with a one-page timeline, printed version history, and the ability to discuss their own essay fluently presents a fundamentally different case than a student who arrives with only indignation — even when the indignation is justified.

What makes AI cases different from plagiarism cases

Here's the part of this that's genuinely new, and it cuts in your favor if you argue it carefully.

A classic plagiarism case has an exhibit: your paragraph next to the source it matches. The fact of the match is checkable by everyone in the room. An AI case usually has no exhibit — there is no source document, because the accusation is that a machine generated original text. What stands in for the exhibit is a probability score from a proprietary classifier. The evidence is probabilistic, and the vendors say so themselves.

Turnitin — the detector behind most of these cases, whose system we've documented in detail on our Turnitin hub — is explicit about the limits. Its headline claims (98% accuracy, under 1% false positives) apply only to documents where more than 20% of the text is flagged. Scores from 1–19% display as an asterisk rather than a number, Turnitin's own acknowledgment that its low-range scores aren't reliable. The company's guidance tells institutions the AI score should not be the sole basis for an adverse action — the human has to make the call. And the false-positive risk isn't evenly distributed: research such as Liang et al. in Patterns (2023) found detectors of this statistical family falsely flagged an average of 61.22% of essays by non-native English speakers, and documented risk runs high for students taught rigid essay structures, technical writers, and neurodivergent writers. Vanderbilt University found the false-positive arithmetic troubling enough to disable Turnitin's AI indicator entirely in August 2023.

None of this means detectors are useless — unedited AI text really is caught most of the time, which is why panels take the scores seriously. But it means a score is the beginning of a question, not the answer to it, and the detector's own vendor agrees. If the entire case against you is a percentage, it is fair and increasingly common to say so plainly: "The instrument's own maker says this number is not proof. Here is my version history, which is."

The flip side of probabilistic evidence: your affirmative evidence matters more than in a plagiarism case, not less. In a match case, the match speaks. In an AI case, the strongest voice in the room can be your documented process. Build the file.

Stages 5–7: Finding, sanction, appeal

The finding arrives in writing: responsible or not responsible (institutions vary in vocabulary). Read the rationale as carefully as the verdict, because appeals usually attack the reasoning, not the conclusion.

Sanctions for a first AI-related finding commonly range from a warning, a zero, or resubmission with a penalty, up through course failure; suspension typically enters the picture for repeat or aggravated cases. Many codes publish sanction guidelines — if yours does and your sanction lands above the guideline for a first offense, proportionality itself can be an appeal ground.

Appeals are narrower than students expect. You generally cannot appeal just because you disagree. The classic recognized grounds are: a procedural error that materially affected the outcome; new evidence that wasn't reasonably available earlier; and a sanction disproportionate to the finding. Appeal windows are short — often 5 to 15 days — and firm. If you're going to appeal, start drafting the day the finding arrives.

When to involve an advisor, ombudsperson, or lawyer

Sooner than feels necessary, is the honest answer.

Academic advisors, integrity advisors, and student-government caseworkers exist at many institutions specifically for this, cost nothing, and know how your particular process actually runs — which panelists ask what, which arguments land. If your school offers one, take the offer at the response-window stage at the latest.

The ombudsperson is a confidential, neutral resource at many universities. They won't advocate for you, but they will explain your options, flag procedural problems, and tell you whether what's happening to you deviates from policy. Because the office is typically confidential and off-the-record, talking to them costs you nothing strategically.

An attorney becomes worth considering when the stakes jump categories: suspension or expulsion on the table, a prior finding, a professional program where a notation threatens licensure (nursing, law, medicine), visa status tied to enrollment, or any suggestion the matter could leave the university's walls. Many codes limit what a lawyer can do in the room, but advice from one outside the room can still shape everything you file. This is also the point to repeat: nothing in this article is legal advice, and a licensed attorney in your jurisdiction is the only source of that.

Tell someone, regardless. These cases are isolating by design of circumstance — you're forbidden to discuss them with classmates in some codes, ashamed to mention them in others. A parent, a mentor, a counselor. The students who handle this process worst are usually the ones who handled it entirely alone.

Your preparation checklist

Before the initial meeting, work through this list. Each item is small; together they are most of a defense.

• Locate and read your institution's academic integrity policy, including the procedure and sanction sections, and note every deadline in your notice.
• Export your document's full version history now, and back it up somewhere the file can't be accidentally overwritten.
• Gather process artifacts: outlines, notes, sources, search history, related messages, writing-center or library records.
• Collect two or three earlier samples of your graded writing, ideally including something written in class.
• Reread your flagged submission until you can discuss any paragraph — its argument, its sources, its choices — without the text in front of you.
• Write a one-page factual timeline of how the assignment was completed, dates included, and stick to it.
• Find out whether you may bring an advisor or support person, and arrange one if you can.
• Check exactly what the course policy said about AI use at the time of the assignment — syllabus wording, LMS announcements, anything in writing — because "what was actually permitted" is sometimes the whole case.

If you were flagged for work you genuinely wrote, our guide on what to do when you're wrongly flagged goes deeper on the evidence side of this, and the false positives page collects the research you may want to cite. Some students also find it useful to see how their own honest writing scores on a public checker before a meeting — a sentence-level readout from a free detector can show you which passages read as machine-typical and why, though the same caution applies to every tool in this category, ours included: no detector score, high or low, proves who wrote anything, and any panel argument should rest on your documented process, not on dueling detectors.

FAQ

What usually happens first when you're reported for AI use? You receive written notice — from the instructor or an integrity office — identifying the assignment, the alleged violation, and the next step, usually an initial meeting. Nothing is decided at this point. Your job is to read the notice carefully, note the deadlines, and start preserving your drafts and history.

Can I be found responsible based only on a Turnitin AI score? Policies vary, but Turnitin itself says the AI score should not be the sole basis for action, and its accuracy claims only apply to documents flagged above 20%. Most institutions treat the score as a starting point that must be weighed with other evidence. If a case against you rests entirely on a percentage, that fact belongs at the center of your response.

What's the best evidence that I wrote my own paper? Document version history showing the work developing over multiple sessions is the strongest single item. Behind it: outlines and notes, search and library records, earlier samples of your writing in the same voice, and your own fluent command of the paper's content when questioned. Panels weigh documented process heavily precisely because detector scores are probabilistic.

Do I need a lawyer for an academic integrity hearing? Usually not for a first, low-stakes case — a campus advisor or ombudsperson is often more useful and free. Consider an attorney when suspension or expulsion is possible, you have a prior finding, you're in a licensure-track program, or your visa depends on enrollment. Many schools restrict lawyers to a non-speaking role, but outside advice can still shape your written submissions. This article isn't legal advice.

What standard of proof applies in these hearings? Most U.S. institutions use the preponderance of the evidence — more likely than not — though some use higher standards like clear and convincing evidence. Your school's code will say. The standard matters: under preponderance, your goal is to make your authorship at least as plausible as the accusation, which documented drafting history does directly.

Should I admit it if I did use AI? Don't lie — invented stories collapse under questioning and convert moderate cases into severe ones. Whether and how to acknowledge what happened is exactly the judgment call an advisor helps with, and many codes reduce sanctions meaningfully for honesty and first offenses. What you should never do is falsely confess to end an uncomfortable meeting, or minimize in ways that contradict evidence you know exists.

On what grounds can I appeal a finding? Typically three: a procedural error that materially affected the outcome, new evidence not reasonably available before, or a disproportionate sanction. Simple disagreement with the decision is usually not a ground. Appeal windows are short — often 5 to 15 days — so start immediately if you intend to file.

How long does the whole process take? It ranges from a week (an informal resolution with the instructor) to a semester or more for contested hearings and appeals. Response windows are commonly measured in days, hearings scheduled in weeks. The variable you control is your own preparation time — which is why preserving evidence on day one matters.

Key facts

• Turnitin's accuracy claims (98% accuracy, <1% false positives) apply only to documents where more than 20% of text is flagged; scores of 1–19% display as an asterisk because Turnitin considers them unreliable (Turnitin AI writing FAQ).
• Turnitin's guidance states the AI score should not be the sole basis for action against a student — the vendor itself frames the score as an indicator, not proof (Turnitin).
• Liang et al., Patterns (2023): seven GPT detectors falsely flagged an average of 61.22% of TOEFL essays by non-native English speakers; 89 of 91 essays were flagged by at least one detector.
• Vanderbilt University disabled Turnitin's AI indicator in August 2023, citing false-positive arithmetic at institutional scale (Vanderbilt).
• OpenAI retired its own AI text classifier in July 2023 after it caught only 26% of AI text and falsely flagged 9% of human writing (OpenAI).
• In Turnitin's first year of AI detection (April 2023–April 2024), 200M+ papers were screened; ~11% showed at least 20% likely AI writing (Turnitin, April 2024).
• Most U.S. institutions decide integrity cases on a preponderance-of-the-evidence standard — "more likely than not" — though individual codes vary and yours controls.

Sources

1. Turnitin — "Turnitin's AI writing detection capabilities FAQs" and AI writing transparency materials (guides.turnitin.com / turnitin.com).
2. Liang, W., et al. — "GPT detectors are biased against non-native English writers," Patterns (Cell Press), 2023.
3. Vanderbilt University — "Guidance on AI detection and why we're disabling Turnitin's AI detector," August 2023.
4. OpenAI — announcement retiring the AI text classifier, July 2023.
5. Turnitin — first-anniversary AI detection data release, April 2024.
6. Washington Post (Geoffrey Fowler) — testing of Turnitin's AI detector, April 2023.