All tools

FREE AI CHECKER FOR DUTCH TEXT

AI checker: was your text written by AI?

This free AI checker scores Dutch text. Paste a text, enter a web page or take a photo, and you see per paragraph how strongly the text reads like AI. It only reads Dutch, which is why the checker below is in Dutch as well.

0 van minimaal 80 woorden

Vizibly bewaart je tekst niet. Voor de beoordeling gaat hij naar de AI-diensten die de check gebruikt.

  1. Plak een tekst, geef een webadres of maak een foto.
  2. De tekst krijgt een score van 0% (mens) tot 100% (AI).
  3. Je ziet waarom de score zo uitkomt en waar het in de tekst zit.

HOW THE CHECK WORKS

What the AI detector looks for

The check looks for patterns that often show up in AI text. One sign on its own says little, because people write like that too sometimes. The score only goes up when several of them come together. These are the ten signs that stand out most in Dutch text, and in your result the check underlines the sentences they appear in.

  1. Em-dashes

    The long dash in the middle of a sentence. Dutch writers rarely use it and put a comma or a period there instead.

  2. "Not X, but Y"

    A contrast nobody asked for, set up only to make the second half sound bigger.

  3. A question with the answer right behind it

    A short question the writer answers right away: "The result? ..."

  4. Everything in threes

    Three adjectives, three examples or three short sentences in a row, again and again.

  5. Stacked "no"s

    "No hassle, no fine print, no surprises": it tells you what is not there.

  6. Typical AI words

    Dutch words like cruciaal, essentieel, naadloos and kortom, the Dutch versions of crucial, essential, seamless and in short.

  7. "From ... to ..."

    A range meant to cover everyone: "from freelancers to multinationals".

  8. Promises with an escape hatch

    "Where possible", "if desired", "if necessary": a promise that softens itself.

  9. Fortune-cookie wisdom

    Sentences that fit any business: "In a world where everything moves faster..."

  10. The summing-up closer

    A paragraph that ends with "In short" or "All in all" and repeats what was already said.

The check also looks for a sentence that announces something important without saying what it is, like "This delivers valuable insights." Which insights, the sentence doesn't say. If a paragraph reads like AI, the check underlines that kind of sentence and tells you what the reader is missing. These sentences don't count toward the score, because people write them too.

The same signs in English text, each with an example and a better version, are in signs of AI writing.

How to read the score

  • 0 to 40%: human

    The text shows signs of a person: uneven sentences, its own choice of words and details only the writer could know.

  • 40 to 60%: could be either

    Some parts read like a person, some like AI. Neat business writing by people often lands around this middle.

  • 60 to 100%: AI

    The sentences are roughly the same length and follow a predictable pattern. What it says could have been about any business, and that's how a language model usually writes.

How reliable the AI checker is

No AI detector knows for sure, and this one doesn't either. A text at 70% wasn't written 70% by AI. It just looks more like AI text than a text at 30%.

I tested the check on 172 Dutch texts. People wrote half of them between 2016 and 2021, before ChatGPT existed. Five different AI models wrote the other half, on the same topics. Of the human texts, 5% scored 60% or higher. Of the AI texts, 86% did. The check runs on Jev, an AI model that doesn't write text but only judges it. What that model can do, you can read in my piece about Jev.

Two things go wrong regularly. Neat business writing by people, like a tight service page, looks like AI. And if you ask a language model to write sloppily and in everyday speech, it often gets through.

So use the result as a pointer. It isn't proof, and certainly no reason to hold anything against anyone.

How Google looks at AI text

Google judges a page on what it gives the reader. Which tool was used to write the text doesn't matter for that. You can read this in Google's guidance on AI content.

There is a line, though. If you use AI to produce pages at scale that are mainly meant to rank higher in Google, that counts as spam under its spam policies. A single text that reads like AI is no problem for Google, as long as it gives your reader something.

What to do with the result

If a paragraph reads like AI, rewrite it in your own words. Add what only you know: a question a customer asked last week, a price, a place name or a mistake you once made yourself. Cut sentences that could apply to any business.

How to tackle a paragraph like that is in signs of AI writing. To see how your pages do in Google, use the keyword rank checker: it shows which search terms your site already shows up for.

OTHER FREE TOOLS

Keep checking

Every tool is free and works without an account.

Website checker

One score for the SEO, speed, security, technical setup and accessibility of a page.

Go to the tool

Keyword rank checker

Up to 10 search terms that fit what you offer, and whether you are already on page 1 of Google.

Go to the tool

SEO checker

Whether Google can read your page properly, and why each point matters.

Go to the tool

Keyword research tool

How often people search for your keywords, up to 10 at once, with the trend over a year.

Go to the tool

Website speed test

How fast your page loads on a phone.

Go to the tool

Accessibility checker

Whether your page works for people with a disability.

Go to the tool

FREQUENTLY ASKED QUESTIONS

FAQ about the AI checker

No. The check is built for Dutch text and tested on Dutch blogs, news articles, web copy, forum posts and reviews. It looks for the patterns language models leave in Dutch, so an English text gives you no reliable score. That is also why the checker itself is in Dutch.

No AI detector knows for sure, and this one doesn't either. The check gives a score from 0% (human) to 100% (AI). Neat business writing by people sometimes looks like AI, and AI text that was deliberately made sloppy sometimes slips through. Use the result as a pointer, not as proof.

Paste the Dutch text into the box: at least 80 words, at most 1,500. The check first scores the whole text and then each paragraph. It looks at the structure, the rhythm of the sentences and whether there are details only the writer could know. Typos don't count, because those are easy to add.

Yes. For a web page, the check reads the running text, without the menu, footer and product cards. For a photo or screenshot, AI first reads the text in the image. You get to see that text, so you can fix anything that was read wrong before the scoring starts.

Google judges a page on what it gives the reader. Which tool was used to write the text doesn't matter for that. According to Google, it can count as spam when you use AI to produce pages at scale that give the reader nothing.

Vizibly doesn't store your text or photo. For the scoring, your text goes to the AI services the check uses. A photo is first made smaller in your browser, without location data, and then read by those same services.

You can do 10 checks a day, up to 3 of them with a photo. A check counts as soon as it starts, including a photo with no readable text on it. If a check fails because the AI service is busy for a moment, you get that check back.