The short answer
- Turnitin is the detector most colleges have access to, and it uses its own in-house AI model, not GPTZero or any third-party tool. Turnitin says it serves more than 16,000 institutions, but its AI indicator is only included for schools that license Turnitin Originality.
- Some colleges license alternatives: the University of Louisiana System uses GPTZero through a data platform; the University of Illinois Springfield and Utah Tech switched to Copyleaks.
- Many universities have AI detection turned off or limited. We verified 16 on their own websites, including Vanderbilt, Yale, Johns Hopkins, Georgetown, Michigan State and UT Austin.
- On admissions essays, the Common App treats submitting AI output as your own work as fraud, but colleges rarely say they run detectors on applications.
- If you’re flagged, ask which tool and score were used, bring your version history and drafts, and point to the school’s own guidance that a score is not proof.
Most colleges that use an AI detector use Turnitin’s, and Turnitin runs its own in-house model rather than GPTZero or another outside tool. But “the college has Turnitin” no longer means “your essay gets an AI score.” Turnitin now includes AI detection only with a specific license, and a long list of universities have switched the feature off or never turned it on. A few schools license GPTZero or Copyleaks instead.
Below is what each tool is, which colleges we could confirm use or reject it (every school checked on its own website), what happens with admissions essays, and what to do if your work is flagged. For how these detectors work under the hood, see how AI detectors work.
What AI detector does Turnitin use?
Its own. Turnitin’s AI writing detection FAQ describes a proprietary deep-learning model based on the transformer architecture, trained on Turnitin’s in-house corpus of academic writing. It is not GPTZero, Copyleaks or OpenAI’s tool. The same model now also flags text it judges was run through AI paraphrasers (added July 2024) or “humanizer” bypass tools (August 2025), in English only.
Scale is why Turnitin dominates the question. Its media kit says it serves more than 71 million students at over 16,000 institutions in more than 185 countries.
The licensing detail matters to students. Turnitin’s FAQ says AI writing detection is only available to customers that license Turnitin Originality; schools with basic Similarity or Feedback Studio packages must buy it separately. CU Boulder’s IT office says keeping the tool after its free trial ended in December 2023 “would have required additional funding,” and the University of Illinois Springfield cited Turnitin’s decision to charge extra for AI detection. Even where it is licensed, only instructors and administrators see the score, and administrators can turn it off.
Which AI detectors do colleges license?
| Tool | What it is | Scale (self-reported) | Example institutions (verified) |
|---|---|---|---|
| Turnitin AI writing indicator | Built into Turnitin; requires Originality license | 16,000+ institutions (all Turnitin products) | Siena College; University of Wisconsin–Superior; University of Melbourne |
| GPTZero | Standalone detector with LMS integrations | 1M+ educators; “used in over 3500 colleges” | University of Louisiana System (via K16 Solutions, 2024) |
| Copyleaks | Plagiarism and AI detector; integrates with Canvas, D2L, Moodle, Blackboard | Not stated for education | University of Illinois Springfield (2024); Utah Tech University (2025) |
| Pangram, Originality.ai | Standalone detectors | n/a | No university license we could verify |
Table: AI detectors colleges license, with institutions confirmed on their own pages or the vendor's named announcement. Checked October 15, 2026.
Scale numbers come from the vendors and count every product, not just AI detection. The institution examples are ones we confirmed: Siena College says it enabled Turnitin’s AI tool on April 4, 2023 and formally adopted it in January 2024; UW–Superior’s Canvas guide covers the AI score; Utah Tech says it “has acquired and implemented Copyleaks,” with AI detection off by default until an instructor switches it on.
Which universities turned AI detection off?
We confirmed 16 on the universities’ own pages: 13 switched it off or dropped it, and 3 limited it or advise against it. The reasons repeat: false positives, bias against non-native English writers, lack of transparency and, since 2024, cost.
| University | Decision (from its own page) | When |
|---|---|---|
| Vanderbilt | Disabled Turnitin’s AI detector “for the foreseeable future” | Aug 2023 |
| University of Pittsburgh | Teaching Center disabled the tool; does not endorse any AI-detection tool | 2023 |
| UCLA | “Temporarily opted out” of the AI detection preview | Apr 2023 |
| Yale | “The AI detection feature of Turnitin is currently disabled” | Undated |
| Johns Hopkins | Disabled Turnitin’s AI detection over false positives | Page updated Mar 2025 |
| Georgetown | Turned the feature off; harms of false positives outweigh benefits | Oct 2023 |
| Michigan State | Turnitin AI detection “not currently available at MSU” | Undated |
| Montclair State | “Officially discontinued in 2023” over reliability and bias | Nov 2023 memo |
| UT Austin | Does not endorse AI detection software; no central contracts with it | Undated |
| University of Minnesota | Does not centrally support or recommend any GenAI detector | Undated |
| CU Boulder | Advisory group unanimously recommended against it; license not upgraded | After Dec 2023 |
| University of Alabama | Removed from Blackboard (still in Feedback Studio) | Aug 2023 |
| Arcadia University | Did not renew the AI add-on over new contract terms | 2026 page |
| University of Waterloo (Canada) | Discontinued Turnitin AI detection | Sep 2025 |
| Curtin University (Australia) | Off from January 1, 2026; text matching stays | Announced Sep 2025 |
| Cornell | Teaching center does not recommend detectors for integrity cases | Guidance only |
Table: universities that disabled, dropped or advise against AI detection, each confirmed on the university's own website. Alabama, Arcadia and Cornell are partial or guidance-only.
A tracker run by GradPilot claims more than 60 institutions have done the same. We list only schools we confirmed directly; Boston University, Northwestern and Oregon State appear in secondary reports but we could not confirm them on their own current pages. The evidence behind these decisions is in our research summary on AI detector false positives.
Do professors use AI detectors on their own?
Some do, and whether they may depends on the school. Where the Turnitin indicator is licensed and on, a professor sees the score automatically when a paper is submitted through the course’s Turnitin assignment. No extra step is needed.
Where it is off, a professor might paste text into a free online detector. Some universities restrict this. UT Austin’s provost office, for example, says third-party detection software is prohibited without a university contract, and the University of Minnesota warns faculty that tools including GPTZero, Copyleaks and Originality.ai cannot provide proof of AI authorship.
K-12 schools are a separate story. Policies vary district by district, and our tracker of AI policies in 32 large US school districts shows how differently they handle AI use and detection. For the teacher’s side of the question, see AI detectors for teachers.
Do colleges use AI detectors on application essays?
Rarely publicly, and the honest answer is that most don’t say. What is on record:
- The Common App fraud policy (last updated August 2023) includes intentionally presenting the substantive content or output of an AI platform as your own work.
- Cornell’s Graduate School says it does not rely on automated AI-detection software as primary evidence of a policy violation, and reviews concerns holistically.
- In 2023 reporting by KQED, the University of San Francisco and USC said they were not using AI-detection software on applications. The University of California said it conducts regular screenings to verify the integrity of responses, without saying how.
The practical takeaway does not depend on detectors: an essay that doesn’t sound like the rest of your application, or that you couldn’t discuss in an interview, is a risk either way. Using AI to brainstorm or check grammar is treated very differently from submitting its output, as our guide to using AI without cheating explains.
What to do if your work is flagged as AI
Respond with process evidence, calmly and quickly. University guidance is remarkably consistent that a score alone is not proof:
| Source | What it says about the score |
|---|---|
| Turnitin | Should not be used as the sole basis for action; Turnitin does not determine misconduct |
| GPTZero | Results should not be used to punish or as the final verdict |
| Michigan State | Detectors should not be the sole basis for adverse actions |
| Johns Hopkins | Instructors should speak to students before making accusations |
| University of Minnesota | A conversation about process is the best way to learn the role of AI |
| UW–Superior | The score is inconclusive; talk with the student first |
Table: official guidance that an AI detector score is not proof of misconduct.
Then work through these steps:
- Ask for specifics in writing: which tool, what score, which passages, and what the course policy on AI says.
- Gather your evidence: version history in Google Docs or Word, drafts, notes, outlines, sources and browser history.
- Offer to explain the work in person. Being able to walk through your argument is strong evidence.
- Quote the school’s own guidance (often on its teaching center or IT page) and Turnitin’s statement that the score isn’t proof.
- Learn the process. Ask the academic integrity office how cases are decided and how you can respond or appeal. Our glossary entry on academic integrity explains the terms.
If you did use AI in a permitted way, say so plainly and show how. Knowing how to cite ChatGPT correctly heads off most of these disputes before they start. Whether detectors are accurate enough to trust at all is covered in our evidence review, do AI detectors actually work?
A caveat on this page: college technology contracts change every year, and many university pages carry no date. We checked every school listed above on October 15, 2026; if your college isn’t listed, its IT help pages or teaching center will usually say whether the Turnitin AI indicator is on.