Find out where the text in a paper comes from.

You have probably put a suspicious sentence in quotation marks and searched for it yourself. That is what we do. On the whole submission, in under a minute, with the source and the student's text shown side by side.

The first paper is free. No account needed, and the paper is not stored.

Found on Wikipedia
Submission, 2,418 wordsExample

In Animal Farm, Orwell shows how a revolution can be betrayed from the inside. His work is characterised by lucid prose, social criticism, opposition to all totalitarianism, and support of democratic socialism. That is exactly what happens on the farm, where the pigs slowly take over.

Appears word for word on en.wikipedia.orgIn the submission without quotation marks or a citation

How it works

We show you what we searched for and what we found, so you can judge for yourself whether there is anything to it.

  1. You upload the submission

    Word, PDF or plain text. The text is read on our server and held only in memory while the analysis runs. Afterwards it is gone.

    Takes a couple of seconds

  2. We pick the sentences worth checking

    Usually 10 to 15 of them: sentences with specific names, dates and facts, the kind that are easy to paste in. Ordinary phrasing that any student would write is skipped.

    About 15 seconds

  3. We search for them word for word

    Each sentence is searched in quotation marks against a web index updated every day. If it exists, you get the page, the link and the matching passage right next to the student's.

    The whole analysis takes under a minute

You get a finding. The judgement is still yours.

The report tells you where the text appears. Whether that is carelessness, a missing citation or deliberate copying is for you to decide. We think that is the right division of labour.

What you get

  • Every sentence found, shown next to its source with the matching passage highlighted
  • A link to the specific page, not just a domain
  • A suggested correct citation you can pass on to the student
  • An assessment of whether the text may have been written with an AI tool, clearly marked as an estimate

What you do not get

  • A percentage that looks like proof
  • Matches from closed archives, such as your institution's own repository
  • Matches from printed books that are not online
  • Rewrites without word-for-word overlap. The AI assessment can hint at those but not prove them

We search the open web as it is today. That is more than any closed database covers, but it is not everything, and you should know that.

About the AI assessment: It is an estimate and gets things wrong both ways. Studies show that texts by students writing in a second language are flagged incorrectly more often. Use it to ask the student a question, not as grounds for an accusation.

Works in the student's language

The document is analysed in the language it is written in, and the search is aimed at sources in that language first. A German essay is searched among German pages, not English ones.

If the student quotes an English novel at length in a Danish essay, we still read the essay as Danish. That sounds small, but it is the difference between finding a Danish essay bank and missing it.

The report comes in the language you read the site in: English, Danish, German or Spanish.

For schools and institutions

You are uploading someone else's text, so here is what happens to it.

The paper is not stored. It exists only in memory while the analysis runs.
The server is in the EU. During the analysis, excerpts are sent to two sub-processors in the US. They are named in the privacy policy.
We are happy to sign a data processing agreement.
There is an API if you want to run checks from your own system. Opgaver.com already does.

Questions we often get

What exactly do you search?

We search the open web word for word, as it stands right now: essay banks, Wikipedia, blogs, news media and public documents. We do not have a database of our own and do not claim to. What we will not find is anything behind a login, or printed books that were never put online.

Can you tell whether a student used ChatGPT?

We give an estimate, and it is labelled as one. It is wrong in both directions, and students writing in a second language are flagged incorrectly more often. Use it to ask a question, not to settle a case. What we can show with certainty is word-for-word text that exists elsewhere.

Which files can I upload?

Word, PDF and plain text. Essays, dissertations, theses, reports. What matters is that there is enough running text to pick sentences from; a page or two is plenty.

What happens to the student's paper?

It is not stored. It sits in memory on our server in the EU while the analysis runs and is discarded once you have your answer. During the analysis, excerpts are sent to two sub-processors in the US, named in the privacy policy. We do not train models on the text.

Can you sign a data processing agreement?

Yes. Write to [email protected] and we will send a draft. We can also enable an API if you want to run checks from your own system.

What does it cost?

The first paper each day is free, with nothing to sign up for. Full access, with every source visible and more papers per day, has not opened yet. You can join the waiting list once you have tried the first one.

Try it on the next submission.

The first paper is free. Nothing to set up, and you have the answer in under a minute.

Check a paper free

One free paper a day. Full access opens later. You can join the list once you have tried the first.

The paper is not stored. During the analysis, excerpts are sent to two sub-processors in the US, named in the privacy policy. We do not train models on the text.