AI Detector

Get a rough heuristic estimate of whether text may be AI-generated. A guide only, not a definitive verdict. No signup, any device.

Remove Ads

Important: This is a rough heuristic estimate based on writing-style signals — not a reliable AI detector and not proof. AI detectors are known to be inaccurate: they can wrongly flag genuine human writing (especially by non-native English writers) and can be easily fooled. Never use this to accuse, grade, or penalize anyone. Treat the result as a loose indication only.

Heuristic AI-pattern score

—/100

Enter some text to analyse

(style-similarity estimate, not a verdict)

Words	0
Sentences	0
Average Sentence Length (words)	—
Sentence-Length Variation (burstiness)	—
Vocabulary Diversity	—
Most-Repeated Phrase Count	0

Higher sentence-length variation and higher vocabulary diversity are more typical of human writing; very uniform sentences and heavy repetition are more typical of some AI output — but both humans and AI produce text across this whole range, so these signals overlap heavily and are not decisive.

Remove Ads

Share on Social Media:

This AI detector gives a rough estimate of whether a piece of text may be AI-generated, directly in your browser. Paste your text to see an estimate. It is a heuristic guide only, not a definitive verdict, and should never be used on its own to accuse anyone or make a decision about a person.

On this page

How to Use the AI Detector Step by Step
What the AI Detector Does and Its Limits
An Estimate Versus Proof
Who Uses an AI Detector
Pro Tips for Using the Estimate Responsibly
Common AI Detector Mistakes to Avoid
Related Tools You May Need Next
Frequently Asked Questions

How to Use the AI Detector Step by Step

Paste your text. Paste the text you want to check into the input box. A longer passage gives the estimate more to work with than a short one.
Run the check. The tool analyses the text against simple heuristics and produces a rough estimate of how AI-like it appears.
Read the estimate. The result is shown as a rough likelihood, indicating whether the text leans towards looking AI-generated or human-written.
Treat it as a guide only. Read the estimate as one weak signal, never as proof. It can be wrong in both directions, so do not rely on it alone.
Never use it to accuse. Do not use the estimate on its own to accuse a person, mark work, or make any decision about someone. It is not reliable enough for that.

Getting a rough estimate of AI-generated text

What the AI Detector Does and Its Limits

An AI detector tries to estimate whether text was written by a person or produced by an AI language model. This tool does so with simple heuristics, looking at surface patterns in the writing, and returns a rough estimate. It is important to understand from the outset that this is a weak, approximate signal, not a reliable test.

AI detection is genuinely hard and no detector is reliable. Detectors produce both false positives, flagging human writing as AI, and false negatives, missing AI writing. Human writing that is plain or formulaic can look AI-like, and AI writing can easily be edited to look human. Because of this, results from any AI detector, including this one, should be treated with great caution.

This is a rough estimate, never proof. AI detection is unreliable and produces both false positives and false negatives. Never use this estimate on its own to accuse anyone, grade work, or make a decision about a person. Treat it only as a weak, informal signal.

This means the estimate is best used as a rough, informal prompt for your own judgement, not as evidence. It can never prove that a specific person did or did not use AI. Using such an estimate to accuse someone, to grade or penalise work, or to make any decision about a person would be unfair and unsound, given how often detectors are wrong.

Everything runs in your browser, so your text is never uploaded and the estimate appears privately on your device. The tool works on any device with no account, but its value is limited to satisfying curiosity and prompting a closer look, always alongside context and judgement rather than in place of them.

An Estimate Versus Proof

Aspect	What the tool gives	How to read it
Estimate	A rough AI-likelihood	A weak signal, not a verdict
False positives	Human text flagged as AI	Common, so be cautious
False negatives	AI text seen as human	Also common
Proper use	Curiosity and a prompt to look closer	Never for accusing or deciding

Who Uses an AI Detector

People curious about a textSomeone runs a passage through the tool out of curiosity, treating the estimate as an informal, unreliable indicator only.

Writers checking their own textA writer checks their own writing to see whether it reads as generic, using the estimate as a loose prompt to add their own voice.

People learning about AI detectionSomeone exploring how AI detection works uses the tool to see how such estimates behave, understanding their limits.

Editors sanity-checkingAn editor uses the estimate as one weak, informal signal among many, never as a basis for a decision about a contributor.

Anyone wanting a rough estimateA person who wants a rough, non-authoritative sense of how AI-like a text reads pastes it in, reading the result with caution.

A reminder that AI detection is only a guide

Pro Tips for Using the Estimate Responsibly

Never accuse anyone based on it. The single most important point: do not use this estimate to accuse a person, mark or penalise work, or decide anything about someone. Detectors are too unreliable to justify that, and doing so can be seriously unfair.

Treat it as one weak signal. At most, read the estimate as one weak, informal signal to be weighed against context, what you know, and your own judgement, never as evidence on its own.

Use longer passages. Very short texts give the estimate little to work with, making it even less reliable. A longer passage gives a slightly more stable estimate, though it remains only a rough guide.

Expect false results in both directions. Plain human writing can look AI-like, and AI writing can be edited to look human. Expect the estimate to be wrong sometimes in both directions, and do not be surprised or place weight on it when it is.

Prefer conversation over detection. If you are concerned about how a piece was written, an open conversation and looking at the wider context are far more reliable and fair than any detector estimate.

Combine with other tools. To understand a text's style, our Readability Score Calculator and Text Complexity Analyzer describe its writing objectively.

Common AI Detector Mistakes to Avoid

Using the estimate to accuse someone. This is the most serious mistake. AI detectors are unreliable and often wrong, so using an estimate to accuse a person of using AI, or to penalise their work, is unfair and unsound. Never do this; the tool is not evidence.

Treating the result as proof. The estimate is a weak, approximate signal, not proof. Treating it as a definitive verdict misunderstands how unreliable AI detection is. No detector can prove who or what wrote a specific text.

Trusting it on short text. Short passages give very unreliable estimates. Reading much into a result from a sentence or two is a mistake, since the tool has too little to work with to give even a rough signal.

Ignoring false positives. Human writing, especially plain or formulaic writing, is often flagged as AI. Ignoring this and assuming a high estimate means AI was used overlooks how common false positives are, and can lead to unfair conclusions.

To understand a text objectively, our Text Readability Score Calculator and Text Complexity Analyzer describe its style, and the Plagiarism Checker checks text against the web.

An AI-likelihood estimate read with caution

Frequently Asked Questions

How does the AI detector work?

It analyses your text against simple heuristics, looking at surface patterns in the writing, and returns a rough estimate of how AI-like it appears. This is a weak, approximate signal, not a reliable test. AI detection is genuinely difficult, and this tool, like all detectors, is often wrong, so the estimate should be read only as an informal, non-authoritative guide, never as proof of how a text was written.

Is the AI detector accurate?

No AI detector is reliable, and this one is no exception. Detectors produce both false positives, flagging human writing as AI, and false negatives, missing AI writing. Human writing that is plain can look AI-like, and AI writing can be edited to look human. Because of this, the estimate is best treated as a rough, weak signal for curiosity, not as an accurate or trustworthy measure.

Can I use it to prove someone used AI?

No. You should never use this tool, or any AI detector, to prove that a person used AI or to accuse or penalise them. Detectors are far too unreliable for that, and doing so can be seriously unfair given how often they are wrong. The estimate cannot prove who or what wrote a specific text. If you have concerns, an open conversation and the wider context are far more reliable and fair.

Why do AI detectors give false results?

AI writing and human writing overlap a great deal, so no simple test cleanly separates them. Plain or formulaic human writing shares features with AI writing and gets flagged, while AI writing can be lightly edited to look human. This is why detectors produce both false positives and false negatives, and why their results, including this tool's estimate, should be treated with great caution.

What should I use the estimate for?

At most, treat it as a weak, informal signal to satisfy curiosity or prompt a closer, human look, always alongside context and judgement. For example, a writer might use it to check whether their own text reads as generic. It is not suitable for grading, accusing, or making any decision about a person, because it is not reliable enough to bear that weight.

Is my text uploaded anywhere?

No. The check happens entirely in your browser, so your text is never sent to a server. The estimate appears on your own device, and nothing you paste leaves it. This keeps the tool private, though its private nature does not make the estimate any more reliable; it remains a rough guide to be read with caution.

Can I use it on my phone?

Yes. The tool runs in your browser and works the same on a phone, tablet or computer, with no app to install. Paste your text and the estimate appears. Because everything runs on your device, it is private on mobile as on desktop, but the same limits apply: the result is a weak estimate, never proof, and should never be used to accuse or decide.

Is the AI detector free?

Yes, it is completely free with no account and no limit. You can check as much text as you like, at no cost. The tool runs in your browser, so there is nothing to install. Remember that being free and easy to use does not make the estimate reliable: it is a rough heuristic guide only, and should be read with caution and never used to make decisions about a person.