If you came looking for Crossplag's AI content detector, the reason you could not find it is that the website no longer exists as a product site. We checked on 29 September 2026: crossplag.com and www.crossplag.com both return a 301 permanent redirect to inspera.com, and the detector's own URL, crossplag.com/ai-content-detector, permanently redirects to Inspera's equivalent page. Crossplag was acquired by Inspera, the Norwegian digital assessment company, in July 2023, and the brand has since been absorbed into it.
That changes what you need. Inspera sells digital assessment and proctoring software to institutions, on procurement cycles, through contracts. It is not a tool an individual student, writer or freelancer signs up for to check a document. So this is not really a comparison page: the honest version is that a product was retired into an enterprise platform, and the useful thing we can add is what Crossplag was actually measured at while it existed, and what to use now.
Reported as HTTP status codes rather than as a narrative, because the status is checkable and a motive is not.
| Address | Status, 29 September 2026 |
|---|---|
| crossplag.com | 301 permanent redirect to www.inspera.com |
| www.crossplag.com | 301 permanent redirect to www.inspera.com |
| crossplag.com/ai-content-detector | 301 permanent redirect to www.inspera.com/ai-content-detector |
| app.crossplag.com | Still returns 200; the legacy application shell loads |
A 301 is a permanent redirect, which is the signal a site uses when a page has moved for good rather than temporarily. The marketing site and the product page are gone. The application host still answers, so existing institutional accounts may still resolve there, but there is no longer a public product page to sign up from.
Inspera announced the acquisition of Crossplag in July 2023. Crossplag was headquartered in Kosovo and sold originality checking together with AI authorship detection. Inspera's stated intention at the time was tighter integration of originality checking and AI writing detection into its digital assessment and proctoring ecosystem, which is what the redirects now reflect: the capability continues to exist inside a larger institutional platform, and the standalone product does not.
We are not going to attribute a motive to the shutdown. No discontinuation announcement for the consumer-facing tool was found, and the observable facts are the acquisition and the redirects. That is what we can support.
Sources: HTTP status and Location headers observed directly on 29 September 2026; Inspera press release announcing the Crossplag acquisition, July 2023. Inspera's own site returns 403 to automated requests, so its current product pages were not read programmatically.
This is the practical reason a Crossplag user needs an alternative rather than a forwarding address.
Inspera is a digital assessment platform: exam delivery, remote proctoring, marking workflow, and now originality and AI writing checks as a component of that. Its customer is a university or examination board, reached through procurement, with contracts, onboarding and administrator configuration. Its technologies are described as in use across 160 countries.
None of that is a criticism. It is a substantial product for the job it does. But if you are a student wanting to check a draft before submitting, a freelancer verifying a subcontractor's copy, or an editor triaging submissions, an enterprise assessment platform is not available to you at any price you would recognise as a price. Crossplag had a self-service path. That path is what closed.
This is the second time this market has done the same thing to individuals, and it is worth naming as a pattern. Turnitin is not sold to individuals either. The tool with the strongest independent showing in the largest academic comparison of detectors is one that the person being assessed by it cannot access. Crossplag moving inside Inspera repeats the shape: detection consolidating into institutional stacks, with the people most affected by a verdict having the least ability to check one.
That asymmetry, rather than any accuracy claim, is the honest reason self-service detectors have a reason to exist. It is also why we publish our benchmark data openly: if the tool assessing you will not show its working, the least a checkable alternative can do is show its own.
Crossplag is one of very few detectors named in a specific peer-reviewed finding, which makes its record unusually concrete.
Liang et al. (2023), published in Patterns, is the study most often cited on detector bias, and Crossplag was one of the tools in it. The seven detectors evaluated were Originality.AI, Quil.org, Sapling, OpenAI's classifier, Crossplag, GPTZero and ZeroGPT. Run over TOEFL essays written by non-native English speakers, they flagged more than 61% on average as AI-generated, against near-zero false positives on essays by native-English US eighth-graders.
Two things need saying about that figure, because it is quoted carelessly everywhere. It is an average across all seven tools, and no per-tool breakdown was published, so it is not "Crossplag's false positive rate". And it is legitimately a measurement that included Crossplag, which is more than can be said for most detectors that have this number attached to them in blog posts.
If you used Crossplag on second-language English writing, that study is the most important thing on this page. The bias it documented is not specific to one vendor; it is a property of reading perplexity and burstiness, because careful second-language English uses a narrower, more consistent vocabulary and that is statistically similar to what these tools look for. Changing tools does not escape it. Knowing about it changes how you read any score, including ours.
We carry the same bias, and we say so on the result rather than in a footnote. More detail in AI detector false positives and why AI detectors are biased against non-native writers.
Source: Liang et al., Patterns 2023. The 61% figure is an average across the seven named detectors; no per-tool split was published, and we do not manufacture one.
Set out as a checklist rather than a sales pitch, because the criteria matter more than which tool you land on.
Self-service, with 3 checks a day free and no account required. Accuracy published as bands rather than a headline figure: 88% to 92% on long-form English of 300 words or more across 15 model families, and 70% to 78% under about 100 words, because short text is where every detector is weakest. A false-positive benchmark of 5.85% over 1,180 academic papers with per-document data downloadable at our benchmark. Sentence-level highlights. English only, stated plainly, with the ESL bias disclosed on the result.
We were not in the Liang study, the Weber-Wulff comparison or the RAID benchmark, and we do not construct a comparable figure for ourselves by analogy to tools that were. Where we sit relative to others is set out in our comparison.
If your institution assesses your work with Inspera, or with Turnitin, a favourable reading from a different detector is not evidence of your innocence. Different models disagree constantly, and a second opinion is context rather than proof. When the problem is an allegation rather than a draft, process evidence does the work: version history, drafts, and a conversation about your argument. How to prove you did not use AI covers what to keep.
Short, and several of them have a one-word answer.
Not as a standalone product. Checked on 29 September 2026, crossplag.com and www.crossplag.com both return a 301 permanent redirect to inspera.com, and crossplag.com/ai-content-detector permanently redirects to Inspera's equivalent page. The application host app.crossplag.com still returns 200 and loads a legacy shell, so existing institutional accounts may still resolve, but there is no public product page to sign up from.
Inspera, a Norwegian digital assessment company, acquired it in July 2023. Crossplag was based in Kosovo and sold originality checking alongside AI authorship detection. Inspera said at the time that it intended to integrate those capabilities into its digital assessment and proctoring ecosystem, which is what the redirects now reflect. We found no separate discontinuation announcement for the consumer-facing tool, so we report the acquisition and the redirect status rather than a motive.
No. Inspera sells digital assessment, proctoring and marking software to institutions through procurement contracts, and it is not a tool a student, writer or editor signs up for to check one document. Crossplag had a self-service path and that is what closed. This mirrors Turnitin, which is also not sold to individuals, so the person being assessed cannot check their own work in the tool assessing it.
Crossplag was one of seven detectors in Liang et al. (2023), published in Patterns, alongside Originality.AI, Quil.org, Sapling, OpenAI's classifier, GPTZero and ZeroGPT. Run over TOEFL essays by non-native English speakers, the seven flagged more than 61% on average as AI-generated, against near-zero on essays by native-English US eighth-graders. That is an average across the group and no per-tool breakdown was published, so it is not a Crossplag-specific false positive rate.
No, and this is the most useful thing to take from the Crossplag research record. The bias is a property of what these tools measure rather than a flaw in one vendor. Detectors read perplexity, how predictable your word choices are, and burstiness, how much that varies. Careful second-language English uses a narrower and more consistent vocabulary, which is statistically similar to generated text. Our detector carries the same bias and we disclose it on the result.
Look for self-service access, published limits rather than only a headline number, a false-positive figure with a named corpus, sentence-level output rather than one document percentage, and an explicit position on second-language English. Our free tier runs 3 checks a day with no account, publishes accuracy as bands (88% to 92% on long-form English of 300 words or more, 70% to 78% under about 100 words), and publishes a 5.85% false-positive rate over 1,180 academic papers with the per-document data downloadable.
It gives you context, not proof. Different detectors disagree with each other constantly, so a favourable reading elsewhere does not establish that the institutional tool was wrong about your document, and no vendor should tell you otherwise. When the issue is an allegation rather than a draft you are still writing, process evidence carries the weight: version history, earlier drafts, notes, and a conversation about your argument.
Also retired, with OpenAI's own withdrawal notice and the 26% and 9% figures behind it.
Read the history →Another repositioning, reported as HTTP status rather than as a motive.
See the status →The other detector individuals cannot buy, and what that asymmetry means in practice.
See the alternative →The Liang finding in full, and which groups carry most of the risk.
Read the guide →1,180 academic papers, 5.85%, per-document data downloadable.
Check our numbers →Where we sit against the detectors that still sell to individuals.
Compare tools →3 checks a day with no account, sentence-level highlights instead of one percentage, accuracy published as bands, and a false-positive benchmark you can download and audit. English only, and we say so.