How Universities Buy Turnitin and AI Detection Tools: $15 Million Investigation

Analysis of 66 US universities reveals procurement patterns for Turnitin and competitors. Public universities must disclose their AI detection plans through RFPs, while private schools negotiate in secret. California alone spent $15+ million, with Turnitin charging some schools 3.6x more than others.

GradPilot TeamNovember 5, 202515 min read
Check Your Essay • Free Daily Review

Turnitin's $15 Million Secret: How Universities Really Buy AI Detection

Want to know if your university is planning to use AI detection? For public schools, the answer is hiding in plain sight—in procurement documents. Private universities? They'll never tell you.

Our investigation into 66 top US universities found just seven active procurement requests for AI detection tools from 2023 to 2025. Meanwhile, Turnitin dominates the market with 71 million students worldwide, despite at least 12 elite universities—including Yale, Johns Hopkins, and Northwestern—disabling Turnitin's AI detection entirely.

Here's what procurement documents reveal about universities' real AI detection plans, why Turnitin costs some schools 360% more than others, and how competitors like Copyleaks and even Pangram Labs are breaking Turnitin's monopoly.

How Procurement Documents Reveal Universities' True AI Detection Plans

Here's something most students and parents don't know: public universities must announce their technology purchases months in advance. These Request for Proposals (RFPs) tell you exactly what AI detection capabilities they're seeking, when they plan to implement them, and sometimes even what they're willing to pay.

Private universities? Complete darkness. Harvard, Yale, Princeton—they negotiate behind closed doors. No public records. No transparency. This creates a massive information gap that vendors exploit.

When Ohio State posts an RFP for "plagiarism/AI detection services," everyone knows they're shopping. Turnitin knows. Copyleaks knows. Competitors can see the requirements and bid accordingly. But when Stanford negotiates? Nobody knows except Stanford and the vendor.

This matters because procurement documents reveal intentions before policies are announced. The University of Texas System's 2022 RFP showed they were evaluating AI detection—but by 2024, UT Austin banned purchasing AI detection tools entirely, citing reliability concerns. The procurement came first. The rejection came later.

The Money Trail: Turnitin's Variable Pricing Exposed

California universities tell the story best. According to investigative reporting by CalMatters and The Markup in June 2025, these institutions spent over $15 million on Turnitin purchases. The California State University System alone will spend $1.1 million in 2025.

But here's where it gets interesting. The price varies wildly:

  • CUNY System: $1.79 per student
  • UC Berkeley: $2.11 per student
  • Cal State System: $2.59-$2.71 per student
  • UC Irvine Continuing Education: $6.50 per student

Same software. Same features. Prices that differ by 360%.

Some schools are getting absolutely hammered. UC Irvine's continuing education division pays nearly four times what CUNY pays. Why? Nobody will say.

The SUNY Buffalo Blueprint: What Universities Actually Want

The most detailed look at what universities want comes from SUNY Buffalo's current request for proposals (RFP UB-2025-00154-0302-RFP-S), due June 23, 2025.

They're demanding a lot:

  • Five-year contract
  • Cloud servers only in continental United States
  • Integration with D2L Brightspace learning management system
  • 24/7 monitoring
  • Full FERPA compliance
  • The university keeps ownership of all student work

That last point matters. Turnitin's business model involves storing student papers permanently in their database. Universities are pushing back, saying student intellectual property shouldn't become corporate assets.

SUNY Buffalo also requires accessibility compliance (WCAG 2.1 Level AA) without overlay technology. They want real accessibility, not band-aids. Plus diversity goals: 17.9% minority-owned and 12.1% women-owned subcontracting.

Ohio State's $105,000 Question

Ohio State renewed with Turnitin in 2024 for approximately $105,000 annually. Their evaluation committee included everyone—Academic Affairs, Libraries, Online Learning, the Drake Institute, multiple colleges, even the Committee on Academic Misconduct.

Five vendors competed. Two made finals. Turnitin won on "usability and privacy/security."

Here's the kicker: Ohio State explicitly required that vendor contracts prohibit using student submissions to train AI models. Universities are realizing their students' work might be training the very AI systems used to police them. The irony isn't lost on anyone.

Why Turnitin Still Dominates Despite Major Flaws

Turnitin processes 200+ million papers across 16,000 institutions. They own the market through strategic acquisitions—buying competitor Unicheck in 2020 just to shut it down in 2023. They bundle plagiarism detection (which works) with AI detection (which doesn't work reliably) and charge extra for the unreliable part.

As we've covered in our analysis of Turnitin's 15% miss rate, the company openly admits to missing roughly 15% of AI-generated text. Yet universities keep paying. Why?

Lock-in. Most schools signed 5-year or 10-year contracts before AI detection existed. Now Turnitin offers AI detection as an "upgrade" for an extra $0.41-$0.48 per student. Schools feel pressure to add it—nobody wants to be the university that didn't try to catch AI cheating.

But the real power is in the database. Turnitin has been collecting student papers since 1998. Every paper submitted becomes part of their comparison database. Universities can't easily switch because they'd lose access to decades of submissions for comparison. It's brilliant. And problematic.

The Great AI Detection Revolt of 2023-2024

Then something remarkable happened. Elite universities started rejecting Turnitin's AI detection entirely.

Universities that disabled AI detection despite paying for it:

  • Yale University ("currently disabled")
  • University of Michigan-Dearborn (April 2023 - made Washington Post headlines)
  • Vanderbilt University (750 false positives per year expected)
  • Johns Hopkins University (false positive concerns)
  • University of Pittsburgh ("not yet reliable enough")
  • Georgetown University (faculty and Honor Council concerns)
  • Northwestern University (accuracy issues)
  • University of Washington
  • University of Oregon ("may not be effective")
  • Oregon State University (disabled September 2024)
  • University of Iowa ("inconsistencies in outcomes")
  • NYU ("not effective enough to license")

Vanderbilt's math is brutal. Even with Turnitin's claimed less-than-1% false positive rate, processing 75,000 papers annually means 750 students falsely accused. That's 750 academic misconduct investigations. 750 potential careers damaged. For false positives.

Universities recommending against use include Penn State ("unreliable"), University of Minnesota ("NOT recommended"), Michigan State ("should not be sole basis for adverse actions"), and University of Virginia (task force recommended "completely prohibiting" in Honor proceedings).

The Texas Warning: When Vendors Disappear

The University of Texas System learned a hard lesson about vendor consolidation. They used Unicheck until Turnitin bought it in 2020. Then Turnitin killed Unicheck in March 2023. Fourteen UT campuses had to scramble for replacements.

UT Austin's response? They now prohibit purchasing AI detection software with procurement cards or personal credit cards, citing student intellectual property and FERPA concerns.

The lesson: Today's competitor is tomorrow's acquisition. And tomorrow's acquisition is next year's discontinued product.

Utah's Power Move: 800,000 Students, One Contract

While individual universities struggle, Utah shows what's possible with collective bargaining. The Utah Education Network covers 800,000+ students across all Utah educational institutions.

They ditched Turnitin for Copyleaks on August 15, 2023. Why? Copyleaks bundles AI detection without extra charges, offers the only enterprise API solution, and supports 100+ languages. One negotiation. Dozens of institutions covered. Massive savings.

E&I Cooperative: The 5,000-School Gorilla

The E&I Cooperative Services RFP #683523 from June 2020 reveals how most universities actually buy these tools. E&I serves 5,000+ institutions—half public, half private. Their five-year contracts with five-year renewal options mean decade-long commitments.

Their evaluation criteria tells you what matters:

  • Pricing: 35 points (highest weight)
  • Terms & Conditions: 20 points
  • Capabilities: 20 points
  • Quality: 15 points
  • Qualifications: 5 points
  • Added Value: 5 points

Money talks, but it's not everything. Technical capabilities and compliance together outweigh cost.

The California Gold Rush: $6 Million and Counting

The Cal State System's spending is eye-opening:

  • 2024 plagiarism detection: $2.59 per student
  • 2025 plagiarism detection: $2.71 per student
  • AI detection add-on (initial offer): $3.05-$3.19 per student
  • AI detection (negotiated): $3.12 per student
  • 2025 total: Over $1.1 million
  • 2019-2025 cumulative: Over $6 million

Some California schools are bleeding money:

  • UC Berkeley: $1.2 million over 10 years
  • College of the Canyons: $500,000+ over 21 years
  • Cal Poly San Luis Obispo: $171,000 from 2020-2024 (then canceled due to low usage)

That Cal Poly cancellation is telling. They spent $171,000 before realizing faculty weren't using it. How many other universities are paying for expensive "shelfware"?

Private Universities: The Invisible Procurement

Here's what our investigation couldn't find: any public procurement records from Harvard, Yale, Princeton, Columbia, Penn, Brown, Dartmouth, Cornell, Duke, Rice, Stanford, USC, Caltech, Emory, Vanderbilt, or Carnegie Mellon.

Why? Private universities don't have to disclose anything. They negotiate in darkness while public universities must show their cards. This information asymmetry means vendors know exactly what Ohio State pays but can keep Stanford's deal secret.

We only know about private university usage through:

  • Policy announcements
  • IT documentation
  • News coverage
  • Legal proceedings (Yale's use of GPTZero emerged in a lawsuit)
  • Student newspapers

The New Competitors: Copyleaks, Pangram Labs, and Others

While Turnitin maintains dominance, cracks are showing. Copyleaks won Utah's 800,000-student contract by bundling AI detection without extra charges. They offer API access, 100+ language support, and simpler pricing.

But the most interesting development? Pangram Labs, initially focused on college admissions essays, now claims detection accuracy that makes Turnitin look outdated. As we detailed in our Pangram Labs coverage, they report near-zero false positives compared to Turnitin's 1% rate that translates to hundreds of false accusations.

Universities evaluating alternatives are looking at:

  • GPTZero: Created by a Princeton student, now used by Yale and others
  • SafeAssign: Blackboard's built-in option, declining as schools leave Blackboard
  • VeriCite: Limited adoption
  • Scribbr, Ouriginal: Smaller players with niche markets

The pattern is clear: Universities want alternatives to Turnitin's monopoly. But switching costs—both financial and operational—keep most locked in.

What Our AI Policy Directory Reveals

Our comprehensive directory of university AI policies tracks 150+ institutions' official stances on AI use. What we've found aligns perfectly with procurement patterns:

  • Schools with active RFPs tend to have stricter AI policies
  • Universities that disabled Turnitin often have more nuanced, educational approaches
  • Private universities (with hidden procurement) have vaguer policies
  • Public universities (with transparent RFPs) have clearer enforcement plans

The correlation is striking: procurement transparency predicts policy clarity. When we can see what tools a university is buying, we can predict how they'll handle AI use. When procurement is hidden, policies stay ambiguous.

Turnitin's Empire: 71 Million Students, Questionable Value

Turnitin dominates with 16,000+ institutions across 185 countries. They've processed 200+ million papers. They own the market through acquisition:

  • 2020: Bought Unicheck, killed it in 2023
  • 2014: Acquired LightSide Labs
  • Owns iThenticate (faculty plagiarism)
  • Owns Gradescope (grading platform)

Their pricing strategy is clever. Base plagiarism detection runs $2.59-$2.71 per student. Want AI detection? That's extra—another $0.41-$0.48 per student. Competitors like Copyleaks bundle everything together.

The Five-Year Lock-In

Almost every contract runs five years. Some run ten. E&I offers five plus five more. Once you're in, you're stuck.

Standard requirements across all contracts:

  • US-based servers only (no offshore backup)
  • University owns all content
  • FERPA compliance mandatory
  • Shibboleth/ADFS authentication
  • LMS integration essential
  • WCAG 2.1 accessibility required
  • 24/7 monitoring

Universities want multi-language support, customizable reporting, and grammar checking. They require vendors to have at least three current higher education clients with 31,000+ students. No startups need apply.

What Triggers New Purchases?

Most universities aren't shopping because they're locked into contracts. New purchases happen when:

  1. Contracts expire (5-year cycles mean 2018 contracts expire in 2023)
  2. Vendors disappear (like Unicheck's discontinuation)
  3. LMS changes (moving from Blackboard to Canvas)
  4. Low usage (Cal Poly's expensive cancellation)
  5. Technology shifts (ChatGPT's November 2022 launch changed everything)
  6. Policy updates (addressing generative AI)
  7. Budget cuts (forcing software evaluation)

The 66 Universities We Investigated

We examined institutions across nine regional clusters:

Ivy League: Harvard, Yale, Princeton, Columbia, Cornell, Penn, Brown, Dartmouth

Big Ten: Michigan, Ohio State, Penn State, Wisconsin-Madison, UIUC, Purdue, Minnesota, Michigan State

California: UCLA, UC Berkeley, UC San Diego, UC Irvine, UC Santa Barbara, UC Davis, Stanford, USC, Caltech

Northeast: SUNY Buffalo, SUNY Stony Brook, CUNY system, NYU, Boston University, Northeastern, Rutgers, UConn, UMass Amherst

South: UT Austin, Texas A&M, Florida, Georgia, UNC Chapel Hill, Duke, Emory, Vanderbilt, Rice

Midwest: Chicago, Northwestern, Washington University, Iowa, Iowa State, Indiana, Missouri, Kansas, Nebraska

West: Washington, Oregon, Oregon State, Arizona State, Arizona, Colorado Boulder, Utah, Nevada

Mid-Atlantic: Georgetown, GWU, Maryland, Johns Hopkins, Virginia, Virginia Tech, Carnegie Mellon, Pittsburgh

State Flagships: South Carolina, Clemson, NC State, Delaware, Vermont, New Hampshire, Rhode Island, Maine

Where We Looked (And Hit Walls)

We searched everywhere:

  • HigherGov.com (found SUNY Buffalo RFP)
  • State procurement portals (20+ states)
  • University procurement sites (40+ universities)
  • Consortium platforms (E&I, Big Ten, NAEP)
  • Paid databases (BidNet, DemandStar, GovWin - all paywalled)

Most detailed RFPs require vendor registration. Budget information stays hidden. Public records requests would be needed for complete contract details.

What This Means for Parents

Your tuition dollars are funding this. At $3 per student annually, a 30,000-student university spends $90,000 yearly. Over five years, that's nearly half a million dollars. For software that many professors won't use and that falsely flags hundreds of innocent students.

Parents should ask:

  • Does your university use AI detection?
  • What's the false positive rate?
  • How are wrongly accused students protected?
  • What happens to student intellectual property?
  • How much is this costing?

What This Means for Journalists

This investigation barely scratches the surface. Every state has procurement records waiting to be analyzed. Every private university has contracts hidden from view. The real story isn't just the money—it's the ethics of universities paying millions for technology they simultaneously condemn as unreliable.

Questions worth pursuing:

  • How many students have been falsely accused?
  • What's the real total spending nationwide?
  • How do vendors justify 360% price variations?
  • Why do universities renew contracts for tools they've disabled?
  • Who profits from student intellectual property?

Reading the Tea Leaves: What Procurement Tells Us About the Future

Procurement documents are crystal balls. They show us what universities will do before they do it. Here's what current RFPs reveal about 2025-2026:

The shift is coming. Universities with contracts expiring in 2025-2026 are explicitly asking for alternatives to Turnitin. They want competitive bids, not sole-source renewals. They're requiring proof of accuracy, demanding transparency about false positive rates, and insisting on ethical data handling.

Public universities lead the exodus. Because their procurement is transparent, we can see public universities actively seeking Turnitin alternatives. SUNY Buffalo's current RFP doesn't mention Turnitin by name but lists them as the previous vendor—a sign they're open to change.

Private universities remain mysterious. Without procurement transparency, we can't predict what Harvard, Yale, or Stanford will do. But when public flagships like Michigan and Berkeley make moves, privates often follow.

The Bottom Line for Students and Parents

Your university's procurement documents tell you more about their AI detection plans than any policy statement. Want to know if your school is planning to use AI detection? Here's how:

  1. Public university? Search "[University name] RFP plagiarism detection" or check your state's procurement portal
  2. Private university? You're out of luck—but watch their peer institutions' public procurements
  3. Check our AI policies directory for current policies at 150+ schools
  4. Look for Turnitin alternatives like Copyleaks or Pangram Labs in procurement documents—it signals dissatisfaction with the status quo

The economics are simple but stark. Turnitin charges universities anywhere from $1.79 to $6.50 per student for the same service. The AI detection add-on costs extra. The false positive rate means hundreds of innocent students face accusations. And universities keep paying because switching costs are too high.

Seven procurement requests. Twelve universities disabling features. $15+ million in California alone. Hundreds of false positives.

This is how universities really buy AI detection. Through opaque processes, variable pricing, multi-year lock-ins, and technology that admittedly misses 15% of AI content while falsely flagging human writing.

The procurement documents don't lie. They show universities know exactly what they're buying: imperfect technology at premium prices. The question isn't whether they'll keep buying it. It's whether students and parents will keep accepting it.


Sources and Methodology

This investigation analyzed:

  • Public procurement databases and RFPs (2023-2025)
  • University policy documents and announcements
  • Investigative reporting from CalMatters and The Markup
  • State procurement portals across 20+ states
  • Vendor documentation and pricing sheets
  • Academic integrity office guidance pages
  • IT services documentation
  • News coverage and legal proceedings

Key limitations: Private universities don't disclose procurement information. Most budget data remains confidential. Vendor registration requirements limit access to detailed specifications.

For students concerned about AI detection and academic integrity, GradPilot provides transparent, ethical writing assistance that helps you develop your authentic voice while meeting university requirements.

Worried About AI Detection?

150+ universities now use AI detection. Check your essays before submission.

Related Articles

Submit Essays That Get Accepted

Join thousands of students who verify their essays pass AI detection before submission

No credit card required

150+

Universities Tracked

99.8%

Detection Accuracy

0.004%

False Positives