CHATGPT DETECTOR · FREE PLAN
ChatGPT Detector: Was This Text Written by ChatGPT?
Paste at least 300 characters. Each passage is scored for AI writing, whichever assistant produced it, and the verdict comes with the reasons, not a bare percentage.
Open in beta · 10 free credits a month · Files deleted after 30 days
AI Text Detector
Open in betaWhat you need
Paste a text of at least 300 characters
At least 300 characters; longer texts give more reliable results.
At a glance
Per passage
Each block of 2–3 sentences gets its own score, so partly generated texts are visible.
AnyFact method
3 levels
Probably human, uncertain or probably AI — never a bare percentage — with the reasons and the limits of the method.
AnyFact method
10 credits
Free every month with an account, no credit card. Uploaded files are deleted after 30 days.
AnyFact free plan
Why a ChatGPT detector should not care which chatbot it was
People say “ChatGPT” for any AI writing, but a text can just as well come from Claude, Gemini, Copilot or an open model. A useful check looks at how the text was written, not at one brand.
Our method measures how predictable each passage is to an open-source language model, the idea behind Fast-DetectGPT. Generated text tends to be more predictable than human writing, passage by passage.
Edited, translated or very short texts are harder to judge, and writing by non-native speakers can be flagged wrongly. The result is an indication to start a conversation, not proof.
Does this sound familiar?
- An email or a cover letter reads like a template.
- An assignment does not sound like the student who handed it in.
- A batch of reviews or comments all share the same tone.
- You want to check your own text before you submit it.
Here's what we check
Paste the text
At least 300 characters; longer texts give more reliable results.
Passage-by-passage scoring
Each block of 2–3 sentences gets its own score, so mixed texts are visible.
A verdict in three levels
Probably human, uncertain or probably AI — never a bare percentage — with a short explanation and the limits of the method.
How to read the result
The signals point to a person. It is not a guarantee: a careful generator or heavy editing can leave few traces.
The signals disagree or are too weak. We say so rather than guess, and list what was found.
The signals point to generation. It is an indication, not proof — check the context before acting.
Sample layout. Real reports name the signals that were checked and the limits of the method.
Can a ChatGPT detector spot Claude and Gemini text too?
Yes, because the method does not depend on any one chatbot. The AnyFact ChatGPT detector measures how predictable a text is to an open-source language model, so it applies to text from ChatGPT, Claude, Gemini or other assistants.
| Question | Answer |
|---|---|
| Does it need to know which chatbot was used? | No. It looks at the text itself, not at a model signature. |
| Does it name the chatbot in the result? | No. It says whether passages look generated, not which assistant wrote them. |
| Are results published per chatbot? | No. We do not publish accuracy figures, per model or overall. |
| What if the text was edited after generation? | Edited passages may score as human, and the text can show up as mixed. |
This matters because many texts pass through more than one tool: drafted with one assistant, polished with another, then edited by hand. A method based on predictability looks at the wording itself, whatever produced it.
How to detect ChatGPT writing, step by step
Run the check, then use the highlighted passages to guide a conversation. The detector tells you where to look; the author can tell you how the text was written.
- 01Paste at least 300 characters in English, French or Arabic.
- 02Read the verdict and open the passages scored as probably AI.
- 03Compare those passages with what you know of the author’s writing: earlier work, emails, notes.
- 04Ask for drafts or version history when the stakes justify it.
- 05Talk with the author. Ask them to explain a highlighted passage or how they found a source.
Schools that do this well set a clear policy first, so students know which uses of an assistant are allowed. Our page for teachers and schools gives a practical process.
Signs of ChatGPT-style text you can check yourself
Some traits are common in chatbot output, but none is reliable on its own. Use them to decide what to ask about, not to decide the answer.
| Sign | Example | Why it is weak alone |
|---|---|---|
| Chatbot leftovers | An opening such as “Certainly! Here is…” or a closing offer to help further | Strong when present, but rare in edited text |
| Textbook structure | Introduction, three balanced points, a summary conclusion | Schools teach exactly this structure |
| Sources that cannot be found | A reference with a plausible title that does not exist | People also misquote and mistype references |
| Even rhythm | Sentences of similar length and shape, paragraph after paragraph | Careful writers can be very regular |
| Sudden change of voice | A text far more fluent than the author’s other work | People improve, and some get legitimate help |
Our guide on how to tell if text is AI-generated goes through these signs with worked examples.
What a GPT detector cannot tell you
A GPT detector estimates whether wording looks generated. It cannot answer the questions that usually matter most after that.
- Whether the use was allowed. Brainstorming, outlining or fixing grammar may be fine under your rules; the detector cannot tell these uses apart.
- Who used the tool. A text can be generated by someone other than the person who submits it.
- Whether the ideas are original. A person can generate wording for ideas that are genuinely theirs, and the reverse.
- Whether the content is true. Human and generated text can both be wrong.
- Whether the author understands the text. A short conversation about it answers that better than any score.
If you are comparing tools, our GPTZero alternative comparison describes another detector and the cases where it may suit you better.
Before you check
Is the result proof that a content was made by AI?▼
No. It is an indication, not proof. Detectors can be wrong in both directions, especially on compressed, cropped, edited or very short content. Use the result to decide what to check next, not as a final judgement on anyone.
Why no percentage?▼
A single percentage suggests a precision no detector has. You get one of three levels — probably human, uncertain or probably AI — with the signals behind it and the limits of the method.
How does the analysis work?▼
When the content carries provenance data, such as C2PA content credentials or metadata, we read it first. Then an open-source detector scores the content. When the score is borderline, Claude reviews it as a second opinion, and the result is explained in plain words.
What happens to my files?▼
They are processed to run your analysis — including, for borderline cases, a review by Claude, Anthropic’s model — and deleted after 30 days.
What does it cost?▼
There is a free plan: a free account gets 10 credits every month, with no credit card. Pro ($15 a month) gives 1,000 credits a month; Business ($49 a month) gives unlimited Business Checks and AI checks under a fair-use limit.
Can I use this result to sanction a student or a candidate?▼
No. Text detectors can be wrong, especially on short texts, heavily edited texts and texts by non-native writers. Use the result to start a conversation, never as the only evidence.
Which languages are supported?▼
English, French and Arabic.
Is a ChatGPT detector different from an AI text detector?▼
Not at AnyFact. The ChatGPT detector is our AI text detector: the same passage-by-passage method applies whichever assistant may have written the text.
Can it detect text from newer ChatGPT versions?▼
The method does not need to recognise a version, because it measures predictability rather than a model signature. Newer models may write in less predictable ways, which is one reason the result is an indication and comes with an uncertain level.
What if someone used ChatGPT only to fix grammar?▼
Light fixes change few words and usually leave the author’s wording in place. Heavier rewriting can make passages score toward AI. Whether that counts as misuse depends on your rules, so ask how the text was produced.
Can I check a PDF or Word file for ChatGPT text?▼
Yes, with the AI document detector. It accepts PDF, DOCX and PPTX files up to 20 MB and 50 pages, and checks the text block by block along with embedded images and metadata.
Check a text for ChatGPT writing, passage by passage
What you need
Paste a text of at least 300 characters
Free plan — 10 credits every month, no credit card.
Each result is an indication, not proof.
Files deleted after 30 days.
You might also want to check
Was This Text Written by AI?
Free AI text detector and AI writing checker: paste a text to see which passages were probably written by AI, with an explained verdict. English, French, Arabic.
AI Essay Checker: Does This Essay Look AI-Written?
AI essay checker for students and teachers: see which passages of an essay look AI-written, with an explained verdict. Paste the text or upload a PDF or Word file.
Was This Document Made With AI?
Check the text, images and metadata of a PDF, DOCX or PPTX file, page by page, with a verdict explained in plain words. Free plan.