LumenWrite
Sign inGet started
Free · No sign-up

Free AI detector and AI checker

Paste your text and see how much of it reads like AI, sentence by sentence. Works on drafts from ChatGPT, Claude, Gemini and other AI writers.

0 / 200 words5 free checks a day, no account needed
Detection score

No score yet

Click “Check for AI” to scan your text.

How the AI detector works

1

Paste your text

Add 40 to 200 words: an essay, an email, a blog post or any AI draft.

2

Check for AI

The detector scores the whole text and every sentence in a few seconds.

3

Read the result

Green sentences read as human, red ones read like AI. The score is the share of the text that reads human.

What the score means

50% human or more: looks human

Most of the text reads like a person wrote it.

Under 50% human: AI detected

Large parts read like AI. Rewrite the red sentences, or let the humanizer do it for you.

No AI detector is 100% accurate. Treat the score as a signal, not as proof, and never use it alone to judge someone's work.

More AI detection tools and guides

ChatGPT detector: find out if ChatGPT wrote it
Free ChatGPT detector: paste 40 to 200 words and see which sentences read like ChatGPT, with a % human score. No sign-up, and your text isn't stored.
AI detector for essays: check your essay before you hand it in
Free AI detector for essays: check your essay in sections of up to 200 words, see which sentences read like AI, and revise before you submit. No sign-up.
AI checker for teachers: a starting point, not a verdict
A free AI checker for teachers: see which sentences in a submission read like AI, where detectors misfire, and how to follow up fairly. No sign-up needed.
How do AI detectors work?
How do AI detectors work? A plain guide to perplexity, burstiness, trained classifiers and watermarks, and why two detectors can disagree on one text.
Are AI detectors accurate?
Are AI detectors accurate? What published studies and detector makers report on false positives, who gets wrongly flagged, and how to use a score fairly.
Why is my writing flagged as AI when I wrote it myself?
Why is your writing flagged as AI when you wrote it yourself? The real causes of false positives, the evidence that shows your process, and next steps.

AI detector FAQ

Yes. You can check 5 texts a day without an account, up to 200 words each. A free LumenWrite account gives you more checks every day, plus the humanizer.

It looks for the patterns language models leave in writing, such as predictable word choices, even sentence rhythm and stock phrases. Every sentence is marked as reading human or AI, and the score is the share of the text that reads human.

It is most reliable on longer, unedited AI drafts and less certain on short or heavily edited text. Different detectors can disagree about the same text, so use the score as a guide, not as proof.

Yes. It checks the writing itself, not which tool produced it, so it works on text from ChatGPT, Claude, Gemini, Copilot and other AI writers.

It works best on English text. Scores for other languages are less reliable.

Short texts don't carry enough signal for a fair verdict. Texts under 40 words aren't scored and don't use up a check.

No. Your text is used only to score it, and we don't save it.

It happens: formal, formulaic writing can read like AI. Vary your sentence length, add specifics and your own voice, or run the text through the LumenWrite humanizer, which keeps your facts and citations.

Turn AI drafts into natural writing

Create a free account to humanize your text, then check it again.

Start free