← Back to blog

Plagiarism and AI at the University of Florida: Draft Coach, version history, and a seven-year record

|10 min read

Check your writing for AI text — free

The first 1,500 words are free, with no sign-up. Every verdict shows how often it is wrong about verified human writing — a figure no other detector publishes.

We are building a writing workspace: your Word or LaTeX document, your PDFs beside it, and an assistant that can only cite what is actually in them — see it and get notified.

The short answer

UF is one of the few universities that lets students run the check themselves before submitting — Draft Coach gives you your own similarity, citation and grammar reports. If it is enabled in your course, use it. It removes the most common way a student ends up in a conduct meeting for something they did not intend.

The second thing worth knowing: UF’s published answer to AI in student writing is not a detection score at all. It is process visibility — a pilot of Turnitin Clarity, where you write inside the platform and your instructor can play back the version history. That changes what protects you. Drafts and revision history are the evidence; a percentage is not.

You can check your own work first

UF e-Learning names Draft Coach in one sentence, and the sentence is the important part:

“Draft Coach™ is a student-facing productproviding immediate access to Similarity, Citation, & Grammar reports.”

Turnitin is centrally supported at UF, through UF e-Learning (learning-support@ufl.edu, 352-392-4357). Most institutions license it for instructors alone, which leaves students guessing about what their own bibliography looks like to a matching engine. UF does not.

Two practical caveats. Whether Draft Coach is switched on for a given assignment is a course-level setting, so it may not be there when you look. And a similarity report answers a different question from the one an AI accusation asks — it tells you what matches existing text, not who wrote yours. That difference is the whole subject of what a Turnitin score actually means.

UF’s answer to AI is version history, not a score

UF is rolling out Turnitin’s Clarity / Student Writing product, described on the e-Learning page as piloted “in Fall 2025”, with an “Enhanced Similarity Report” listed as “Available Spring 2026”. The tool finder run by Learning Support Services describes Clarity itself — the page carries no date, which we note rather than paper over:

“Turnitin Clarity is a student-writing tool that combines plagiarism detection with real-time AI writing assistance in an integrated composition space.”

“Currently in an extended pilot and available for faculty to enable it in their classes.”

“Campus Licensed. UF covers the cost for usage.”

Inside it, students receive “AI-powered feedback, grammar suggestions, and citation help”, while instructors gain visibility into “student progress, AI usage, and originality scores throughout the drafting process” by way of version-history playback.

Read that carefully, because it cuts both ways. If your course uses Clarity, your instructor can watch the document grow. A thesis chapter that appears in one paste looks different from one that accumulates over three weeks — and that is the point of the design.

The defence this creates is the same thing that makes it uncomfortable: a documented writing process. Whether or not your course uses Clarity, keeping dated drafts, outlines and notes is the single most durable answer to an accusation. We wrote up how to build that record in how to prove you wrote it yourself.

Is the AI indicator switched on? UF does not say

This is a documented absence, and it deserves a plain sentence rather than a guess. UF does not publish whether Turnitin’s AI-writing indicator is enabled or disabled. Its e-Learning Turnitin page does not mention AI writing detection at all — not to turn it on, not to turn it off. Searches restricted to UF’s e-Learning and Learning Support Services domains returned no page containing the phrase “AI detection”. No statement from UF e-Learning or the Center for Teaching Excellence establishing a position could be located.

Silence is not the same as “on”, and it is not the same as “off”. It means the question has no published answer, and anyone who tells you otherwise about UF is inferring.

Note too that Clarity, which UF does document, is not the AI-writing-detection indicator. They are separate products answering separate questions: one shows how a document was written, the other prints a number about a finished file.

When a syllabus claims AI detection

Here is something you will run into, and it is worth understanding before it worries you. Individual UF course syllabi published on ufl.edu subdomains do assert AI detection. One Anthropology syllabus states:

“Assignments will use the tool Turnitin to assess plagiarism and are subject to ChatGPT/AI detection.”

That is one instructor’s course document. It is not a university position, and there is no published UF policy behind it. We say so because the gap matters to you in two directions:

  • It still binds your conduct in that course. The syllabus is what sets what is authorized. If it says AI use is not permitted, that is the rule you are graded against, whatever the detection sentence means technically.
  • It is not evidence of a university-wide capability. If you are ever told “UF detects AI”, the honest answer is that UF has published no such position, and a syllabus sentence is not one.

The reasonable move when a syllabus is vague or sweeping is to ask the instructor what specifically is permitted, in writing, and keep the reply. It costs a minute and converts an argument into a document.

What Regulation 4.040 actually prohibits

The governing document is UF Regulation 4.040 — Student Honor Code and Student Conduct Code. One honest note first: no adoption date, amendment date or approving-body signature appeared in the text of the regulation as published at policy.ufl.edu, so we do not attribute a date or a body to it here.

The pledge is one sentence:

“On my honor, I have neither given nor received unauthorized aid in doing this assignment.”

Plagiarism is defined so that students must not represent the work of another person or Entity as their own, and the definition explicitly reaches “stealing, misquoting, insufficiently paraphrasing, or patch-writing” — and “self-plagiarism, which is the reuse of the Student’s own submitted work.”

Two of those catch honest students regularly. Patch-writing — stitching a paragraph together from source phrasing with your own connective tissue — is named as plagiarism, not as a lesser sin. And self-plagiarism means the seminar paper you are proud of cannot simply reappear inside your thesis without permission.

Artificial intelligence is not named anywhere in the violation definitions. The nearest operative language is technology-neutral, and it is what an AI case at UF would have to run on: cheating includes “Using unauthorized materials without consent or proper attribution” and using “resources Faculty hasn’t expressly permitted”.

The word doing the work there is unauthorized. Authorization comes from your instructor. That is why the syllabus, and the emailed answer when the syllabus is silent, is worth more to you than any opinion about detectors.

There is one further clause worth reading with your own eyes. A separate violation, Submission of Purchased Academic Work, covers “academic work in any form generated by an Entity; academic materials in any form prepared by a commercial or individual vendor”. “Entity” is a defined term in the regulation, broad enough to reach a service. UF does not say this clause applies to generative AI, and we are not asserting that it does — but you should know it is there.

The full list of violations on the page: Cheating; Complicity; False or Misleading Information; Interference with Academic Activity; Plagiarism; Submission of Purchased Academic Work; Unauthorized Recordings; Unauthorized Taking or Receipt of Materials.

On disclosure: no university-wide AI-disclosure form or required citation format for AI use was located on UF policy or e-Learning pages. There is no UF template to follow, which means the format your instructor asks for is the format. If you need a starting point, our AI disclosure statement guide has wording you can adapt.

Where your submitted file goes

Partly documented, and the part that is documented is instructor-facing. UF e-Learning states that with the external tool, instructors may elect “to not store submissions in the repository.”

What is not published: whether storage is the default, and how long a stored submission is retained. No student-facing opt-out is described anywhere we could find. So if it matters to you — and for an unpublished thesis it might — the person who can answer is the instructor who configured the assignment.

The record: seven years, and an overlay

UF does not publish an “XF”-style grade notation of the kind some universities use. What it publishes instead is a record and a transcript overlay.

  • Student conduct records are maintained a minimum of seven years.
  • Service Indicators are placed on records for separation, charges, failures to appear, or incomplete sanctions. Note that charges is in that list — a Service Indicator can exist while a case is still open.
  • Transcripts are issued with an overlay stating the student “is not in good standing with the University.”

How long that overlay persists beyond the sanction is not publicly documented. We would rather say so than invent a term.

The published sanction categories are:

  • Academic Sanctions: Grade Adjustment; Drop Restriction.
  • Status Sanctions: written reprimand; conduct review; probation; loss of University privileges; deferred suspension; suspension; expulsion.
  • Educational Sanctions: seminars, assignments, substance consultation; restitution and community service; residence hall transfer or removal; no contact directive.

“Drop Restriction” is the one students overlook: withdrawing from the course is not necessarily available to you once a case exists.

UF publishes no annual report of honor-code case counts or outcomes that we could locate. So nobody — including us — can tell you how often any of this happens.

Proof, hearings and the ten-day appeal

Charges are brought by the “Director of Student Conduct and Conflict Resolution or designee”. Decisions are made by individual Honor Hearing Officers; Student Conduct Committees of three to five members, at least half of them students; University Officials Boards of three faculty or University Officials; or Health Science Conduct Committees.

The routes run from an Information Meeting to one of: Administrative Review, Restorative Dialogue, or a Hearing. The first two are available only where the student accepts responsibility and Separation is not a possible outcome. That is a real fork. If you dispute the allegation, the hearing is your route, and accepting responsibility to make it end faster is a decision with consequences.

The standard of proof is stated in three words:

Preponderance of the Information

Meaning: more likely than not. It is the civil standard, not the criminal one — but it is a standard the university has to meet, and a detector percentage on its own is a thin thing to meet it with. If you are in this position now, read what to do when you are accused of using AI before your meeting, not after.

If a detector score is what started this, our guide to an AI detector flagging your thesis sets out what that number does and does not establish on its own.

Appeals are written and due within 10 Class Days. That is the deadline to diary the moment you receive an outcome. The grounds are:

  • a procedural rights violation that materially affected the outcome;
  • new relevant material information that was unavailable at the hearing;
  • inappropriate sanctions given the violation and the student’s history.

Appeals are decided by the Vice President of Student Life or designee; where Separation or Sexual Misconduct is involved, by an appeal panel of one student and two faculty or staff members.

If English is not your first language, the specific research on false-positive rates for non-native writers is worth having in hand before a meeting — see AI detectors and non-native speakers, and how accurate AI detectors really are for what the numbers do and do not show.

Before you submit

  • Run Draft Coach if your course offers it. It is student-facing and gives you the similarity and citation reports directly.
  • Keep dated drafts and version history.UF’s own direction of travel — Clarity, version-history playback — is toward process evidence. Have some.
  • Get the AI rule for each course in writing, and keep the reply. “Unauthorized” is the word the regulation turns on, and only your instructor can authorize.
  • Check your citations exist and say what you claim. Fabricated references are the fastest route from a suspicion to a finding — see how to verify your citations.
  • Do not reuse your own earlier submitted work without permission. Regulation 4.040 names self-plagiarism explicitly.
  • If an outcome arrives, count ten Class Days. That is your appeal window, and it is short.

If you want a read on your own text before you hand it in, our AI check is free up to 1,500 words, with no account and no name. Our published error rates are measured on English and German corpora, and no detector — ours included — produces proof. What proves your authorship is your drafts.

Sources

This page is orientation, not legal advice. What binds you is UF Regulation 4.040 and your course syllabus.

Frequently Asked Questions

Can UF students check their own work for similarity?

Yes. UF e-Learning lists Draft Coach as “a student-facing product providing immediate access to Similarity, Citation, & Grammar reports.” It is the student-side route, and it is unusual — most universities license Turnitin for instructors only. Whether it is enabled in a particular course is set by the instructor.

Does the University of Florida use AI detection?

Not documented either way. UF's e-Learning Turnitin page does not mention AI writing detection at all — it neither enables nor disables it, and no statement from UF e-Learning or the Center for Teaching Excellence establishing a position could be found. Individual course syllabi published on ufl.edu subdomains do assert AI detection, but a syllabus is one instructor's document, not a university position.

Does UF's honor code mention artificial intelligence?

No. “Artificial intelligence” does not appear in the violation definitions in UF Regulation 4.040. The operative language is technology-neutral: cheating covers “Using unauthorized materials without consent or proper attribution” and using “resources Faculty hasn't expressly permitted.” Whether an AI tool is unauthorized is therefore decided by your instructor, not by the regulation.

How long does a UF conduct record last?

Student conduct records are maintained for a minimum of seven years. UF does not publish an “XF”-style grade notation; instead, transcripts are issued with an overlay stating the student “is not in good standing with the University,” and Service Indicators are placed on records for separation, charges, failures to appear, or incomplete sanctions. How long the overlay persists after the sanction ends is not published.

Check your writing for AI text — free

The first 1,500 words are free, with no sign-up. Every verdict shows how often it is wrong about verified human writing — a figure no other detector publishes.

We are building a writing workspace: your Word or LaTeX document, your PDFs beside it, and an assistant that can only cite what is actually in them — see it and get notified.