Skip to content
ebbmark
Free AI detector · No sign-up

Is this text written by AI?

Paste it below: our free AI text detector answers in seconds and shows which AI it sounds like.

0 / 1,000 words

  • 10 free checks a day
  • 50 to 1,000 words

How it works

Three quick steps

ebbmark is a free online AI detector: an AI writing and content checker for English essays, articles, emails and posts. No sign-up, nothing to install.

1. Paste your text

Any English text from 50 to 1,000 words: an essay, an article, an email or a post.

2. We check it

A detection model reads the text in a few seconds.

3. Get a clear answer

A plain verdict, how sure it is, the percentages and which AI it sounds like.

The detection model’s published results for model 1.0.0: When the model gives a clear answer, AI or human, it is right more than 98% of the time. How this was measured

How ebbmark works, in detail

Which AI wrote it?

Find out which AI it sounds like

When a text reads as AI-written, ebbmark also shows the model family it sounds most like. It’s a resemblance in writing style, not proof of the tool that was used.

Not sure? Then we say so: “AI-written, the model family is unclear”.

More about which AI wrote a text Checking for ChatGPT? Use the ChatGPT detector

  • ChatGPT / OpenAI
  • Claude / Anthropic
  • Gemini / Google
  • DeepSeek
  • Other (open-weight models)

FAQ

Good questions

Is ebbmark free?

Yes. There’s no sign-up and no payment. You can run 10 checks a day, each with 50 to 1,000 words of English text. The site is funded by ads.

What do the results mean?

Every analysed text gets one of four answers. AI-written: the text reads like it was written by an AI model. Human-written, edited with AI: a person wrote it, and an AI tool edited or polished it. Uncertain: the signals are mixed, so the result doesn’t point either way. Human-written: the text reads like a person wrote it, without AI editing.

Can ebbmark tell which AI wrote a text?

For texts that come back AI-written, ebbmark shows the model family the writing sounds most like: ChatGPT / OpenAI, Claude / Anthropic, Gemini / Google, DeepSeek or Other (open-weight models). It’s a resemblance in style, not proof of the tool that was used.

Is a result proof that AI was used?

No. A result is an estimate made by a statistical model, and it can be wrong in both directions. Use it as one input among several, and don’t base a decision about a person on the result alone.

See all questions