Academic Integrity Platforms in 2026: What Changed, and What Your Renewal Checklist Should Say

If your integrity platform comparison spreadsheet was built more than a year ago, at least one column in it describes a product that no longer sells and at least one describes a product that has changed category. That is the state of this market, and it is the reason a renewal review should begin by verifying existence rather than comparing features.

What follows is what we could establish from vendors’ own pages in August 2026, what we deliberately could not, and the checklist we would take into a renewal.

Why “which is best” is the wrong opening question

Integrity platforms are not interchangeable products differing by quality. They occupy at least three distinct categories that institutions routinely conflate: similarity screening, which matches submitted text against a corpus; authorship inference, which estimates whether text was machine-generated; and process visibility, which records how a document came to exist. A ranking that mixes all three answers no question, because an institution short of process evidence is not helped by a better matcher.

So the useful first move in a renewal is to name which of the three you are actually short of. Almost every institution we speak to has screening and lacks process evidence — and then runs a procurement exercise scoring vendors on screening. The specification gets written from the last cycle’s specification, and the gap survives another contract term.

What actually changed

Ouriginal has been withdrawn

Turnitin’s own page states plainly that the Ouriginal service ended on 30 June 2026. It thanks visitors for their interest and directs them to Turnitin Similarity, which it describes as offering “many of the same features as Ouriginal, including comprehensive Similarity Reports with percentage matches and source lists, seamless LMS integration, and full technical and training support.”

This matters beyond housekeeping. Ouriginal appears in a great deal of comparison content, in institutional procurement templates, and in tender documents as a named alternative supplier. A tender that lists a withdrawn product as a comparator is weaker for it, and any market-testing exercise that counted it as an independent option was counting a product that is now part of the incumbent’s portfolio.

There is a structural consequence too. When a comparator is absorbed rather than merely failing, the number of genuinely independent suppliers in a category falls, and with it the institution’s negotiating position at renewal. If your procurement rules require evidence of market testing, it is worth establishing early — while you still have a year of contract left — whether the alternatives on your list are independent suppliers or portfolio siblings.

The category has moved from detection to process visibility

Turnitin’s current product overview leads with Turnitin Clarity, presented as “Don’t just detect AI. Understand it,” and described as bringing writing analytics and AI insights together to provide “the visibility needed to guide responsible AI use and empower authentic learning journeys.” The wider line-up is Feedback Studio, Gradescope, ExamSoft, Similarity and iThenticate.

Read as a procurement signal rather than marketing, that is the incumbent conceding the argument that detection alone does not answer the institution’s question. If your renewal is being justified internally on detection capability, note that the vendor’s own emphasis has moved.

One vendor sells both detection and evasion

This one is worth stating carefully and without embellishment, because it is simply what the vendor publishes. Grammarly’s own site navigation lists, among its AI writing tools, both an AI Detector and an AI Humanizer, alongside a Plagiarism Checker, Paraphrasing Tool, Citation Generator and an AI Grader agent.

An “AI humanizer” is a tool whose purpose is to make machine-generated text less identifiable as machine-generated. An institution can reasonably conclude that this is a normal consequence of serving both consumer and institutional markets. It can also reasonably conclude that a supplier selling both sides of the integrity problem is a supplier whose institutional and consumer incentives point in different directions. Either conclusion is defensible; being unaware of it during a procurement is not.

The practical step is small: run your portfolio check across the vendor’s whole catalogue rather than the products in your tender lot, and record what you found. It takes ten minutes and it is the kind of diligence a governing body expects to see evidenced rather than assumed.

What we did not verify, and why we are saying so

We attempted to open Copyleaks and Turnitin’s AI-detection documentation directly and received automated bot challenges rather than content. Writefull’s institutional page returned a not-found error. We therefore make no claims about those products’ current capabilities, accuracy, pricing or terms.

We are flagging this rather than filling the gap from secondary sources because the failure mode in this category is specific and expensive: comparison articles recycle accuracy percentages that originate in vendor marketing, and those numbers then appear in institutional business cases as though they were independent findings. Once a figure has been through two intermediaries it carries the authority of a citation and none of the evidence of one. If a number matters to your decision, it should come from a document you can produce in a committee.

The renewal checklist

Row three is the one to run first, because it converts an abstract quality debate into a workload number your integrity office will recognise. The arithmetic and the published example behind it are in our analysis of whether AI detection is reliable enough to base a case on.

Row ten is not rhetorical. Ouriginal’s customers received an end date. Continuity planning for an integrity platform is now a live requirement rather than a theoretical one, and the answer should name the fallback, the data you would need to export, and how long a transition would take mid-session.

Detection and support are different purchases

A distinction worth holding through the review: screening tools examine work after submission, and support tools change how work is produced. They answer different questions and they are not substitutes.

Most institutions need both, and the honest framing for anyone selling into this market — ourselves included — is complementarity rather than replacement. An institution with a functioning similarity screening contract should not be asked to abandon it. What it may be missing is visibility into the writing process, which is precisely the gap the incumbent’s own new product positioning acknowledges.

The budget consequence is worth naming plainly, because it is the argument a finance committee will make anyway. If the two are complements rather than substitutes, then adopting process visibility is an additional line rather than a saving, and a business case that pretends otherwise will be found out in year two. The defensible case rests on caseload and staff time rather than on licence displacement.

If you would like to work through where that gap sits at your institution before your renewal date, request an institutional evaluation.

How to sequence a renewal

1. Twelve months out: restate the problem in one sentence with a number. Do not open last cycle’s specification yet.
2. Nine months out: verify which products exist and issue the written question set to each — see our procurement question bank.
3. Six months out: run data protection review in parallel, using the sequence in our data protection guide.
4. Four months out: pilot the leading candidate in one department with pre-agreed criteria.
5. Two months out: decision paper against the criteria, including the do-nothing option costed.

The do-nothing option is worth costing properly. Sometimes the correct answer to a renewal is to renew the screening contract unchanged and spend the marginal budget on assessment redesign, which no vendor will propose to you.

Frequently asked questions

Is Ouriginal still available?

No. Turnitin’s page states the service ended on 30 June 2026 and directs former users to Turnitin Similarity.

What replaced it?

Turnitin Similarity, which the vendor says offers many of the same features including similarity reports with percentage matches and source lists, LMS integration, and technical and training support.

What is in Turnitin’s current line-up?

Turnitin Clarity, Feedback Studio, Gradescope, ExamSoft, Similarity and iThenticate, per its own product overview.

Does Grammarly sell an AI detector?

Its site lists an AI Detector among its AI writing tools, and also lists an AI Humanizer.

Why will you not rank the detectors by accuracy?

Because we could not open the vendors’ own current documentation, and published accuracy figures in this category generally originate with vendors. We will not launder a marketing number into a procurement input.

Is iThenticate different from Turnitin Similarity?

They are listed as separate products in the same portfolio, with iThenticate positioned for publishers, researchers and scholars.

Should we replace our detection contract?

Not as a default. Establish what your policy allows the output to prove, then decide whether the gap you have is screening or process visibility.

How do we compare on data protection?

Through the data processing agreement and sub-processor list rather than the trust page — the approach set out in our data protection review guide.

What if a vendor will not answer in writing?

Treat that as the answer, and record it. Responsiveness during a sales cycle is the best available signal of responsiveness during a contract.

How often should we review?

Annually, given that a product in the standard comparison set was withdrawn this year.

Who should own the review?

An academic owner accountable for integrity outcomes, with procurement and the data protection officer supporting — not procurement alone, or the specification will be written from last cycle’s document.