AI Detector
Check if a text was written by ChatGPT, Gemini, Claude or another AI. Free, instant and private.
Is it AI?
Paste at least 80 words and press Check for AI. You get one of three answers:
- Probably AI
- Can't tell
- Probably not AI
No text to hand? Try a sample: or .
- Free and unlimited
- No sign-up
- Text never leaves your browser
- Tested on 20 AI models
Free AI detector for ChatGPT, Gemini and Claude text
Probably Not AI is a free AI detector and AI checker. Paste an essay, an email, an article or a social post, and it tells you in seconds whether the text reads as AI-generated or human-written. It works on text from ChatGPT, Gemini, Claude, DeepSeek, Grok and other AI writing tools.
The AI content detector gives a plain result, Probably AI, Can't tell or Probably not AI, with an AI score out of 100. There is no sign-up and no word limit, and your text is checked in your browser, so nothing is uploaded.
How to check if text was written by AI
- 01
Paste your text
Paste or type the text you want to check into the box. The AI detector needs at least 80 words.
- 02
Click Check for AI
The AI checker scans the text in your browser. Nothing is uploaded, and the result appears in about a second.
- 03
Read the result
You get a clear answer, Probably AI, Can't tell or Probably not AI, and an AI score out of 100.
What is an AI detector?
An AI detector, also called an AI checker, AI content detector or ChatGPT detector, is a tool that estimates whether a text was written by a person or generated by an AI model such as ChatGPT, Gemini or Claude. It does this by analysing how the text is written: the word choice, the punctuation and the rhythm of the sentences.
People use an AI writing detector to check essays and assignments, to review articles and submissions before publishing, and to see how their own writing will be read. An AI detector gives an estimate. It cannot prove who wrote a text.
How does the AI detector work?
AI models write in recognisable ways. They favour certain words and phrases, they punctuate differently from someone typing on a keyboard, and they keep their sentences more even in length. The AI detector scans your text for these signs of AI writing and weighs them against the signs of human writing, such as typing slips and uneven rhythm.
It then compares the result with 3,379 human and AI texts we tested, and gives you an AI score and a result. See how the AI detector works.
How accurate is the AI detector?
- 95%
- of unedited chatbot text flagged as Probably AI
- 1 in 30
- human texts wrongly flagged as Probably AI
- 3,379
- test texts from people and 20 AI models
We tested the AI detector on essays, emails, articles, forum answers and research writing. It is most accurate on text pasted straight from a chatbot. Across all AI text in the test, including formal and edited writing, it flagged 59%. AI text that has been rewritten or humanized is the hardest to detect.
No AI detector is 100% accurate, and false positives happen with every tool. Use the result as an indication, not as proof. Read the full AI detector accuracy test.
Why use this AI checker
- 01
Free AI checker with no limits
Check as many texts as you want. There are no credits, no daily cap and no premium tier.
- 02
No sign-up
You do not need an account or an email address. Open the page, paste your text and check it for AI.
- 03
Private by design
The AI detection runs in your browser. Your essay, email or article is never uploaded and never stored.
- 04
A clear answer
No confusing percentages. The AI content detector gives one plain result and an AI score out of 100.
- 05
Instant results
The check takes about a second, on a phone, a tablet or a computer.
- 06
Honest results
You get Probably AI, Can't tell or Probably not AI. When the text gives too little to go on, the AI checker says so.
Who uses an AI detector?
Students
Check an essay or assignment before you hand it in, and see how an AI detector reads it.
Teachers and educators
Get a quick first read on student work, and a reason to look closer or ask a question.
Writers and editors
Check drafts, guest posts and freelance submissions for AI-generated text before they are published.
SEO and content teams
Review AI-assisted articles and outsourced content before they are published.
Recruiters
Check cover letters and written answers when you want to know whether a candidate wrote them.
Anyone who is curious
Got an email, a review or a social post that sounds off? Paste it in and check if it is AI.
Which AI models can it detect?
The AI detector checks text from any AI writing tool. It has been tested on text from 20 current models, including:
- OpenAI GPT (ChatGPT)
- Google Gemini
- Anthropic Claude
- DeepSeek
- xAI Grok
- Qwen
- Kimi
- GLM
- Nemotron
Detection rates differ from model to model and depend on how the text was prompted and edited. The accuracy page lists the result for every model in the test.
AI detector vs plagiarism checker
A plagiarism checker compares a text with websites, books and papers to find copied passages. An AI detector looks at the writing itself to estimate whether a person or an AI model produced it. AI-generated text is usually new, so a plagiarism checker will not flag it. If you need to know both whether a text is original and whether it is AI-written, you need both checks.
AI detector FAQ
Is this AI detector free?
Yes. The AI detector is completely free. There is no sign-up, no daily limit and no word limit, and you can check as many texts as you like.
How do I check if a text was written by AI?
Paste the text into the box and press "Check for AI". In a second you get one of three results (Probably AI, Can't tell or Probably not AI), and an AI score out of 100.
Can it detect ChatGPT?
Yes. The AI checker works on text from any AI model, including ChatGPT. It is most reliable on text pasted straight from a chatbot: in our tests it flagged 95% of unedited chatbot emails, essays and posts. On formal, academic-style text from OpenAI's GPT models it flagged about 6 in 10, and edited text is harder again.
Does it work on Gemini, Claude, DeepSeek and Grok?
Yes. We tested the AI detector on text from 20 AI models, including Google Gemini, Anthropic Claude, OpenAI GPT, DeepSeek, xAI Grok, Qwen and Kimi. Results differ by model and by how the text was prompted, and the full table is on the accuracy page.
How accurate is the AI detector?
In our tests on 3,379 texts, it gave the result "Probably AI" to 59% of AI texts and to about 1 in 30 human texts. For unedited chatbot output the rate was 95%. No AI detector is 100% accurate, so treat a result as an indication, not a verdict.
Can AI detectors be wrong?
Yes. Every AI detector makes two kinds of mistake: it can flag human writing as AI (a false positive), and it can miss AI writing, especially after editing. That is why our results say "probably", and why a result should never be the only reason to accuse someone of using AI.
Why was my own writing flagged as AI?
Careful, formal writing shares many traits with AI writing: polished vocabulary, even sentences and lists of three. About 1 in 30 human texts in our tests got the result "Probably AI". If that happens to you, your drafts and version history are the best proof that the work is yours.
Is my text saved or uploaded?
No. The AI check runs entirely in your browser. Your text is never sent to a server, never stored and never used to train anything.
Is there a word limit?
There is no maximum. You need at least 80 words for a result, and longer texts give a more reliable one.
Which languages does the AI checker support?
English only for now. The AI detector was built and tested on English text, and results for other languages are not reliable.
Can teachers use it to check student work?
Yes, as a first look. A result of Probably AI is a good reason to talk to the student and ask about the work. It is not proof of cheating, and no AI detector should be used as the only evidence.
Can it detect AI text that has been edited or humanized?
Sometimes. Editing removes many of the signs an AI detector looks for. When we stripped the most obvious marks from AI texts, the share flagged as "Probably AI" fell from 59% to 45%.
Is an AI detector the same as a plagiarism checker?
No. A plagiarism checker compares a text with existing sources to find copied passages. An AI detector looks at how the text is written to estimate whether a person or an AI model wrote it. AI-generated text is usually new, so a plagiarism checker will not catch it.
Does an em dash mean a text was written by AI?
Not by itself. In our tests an em dash appeared in 65% of unedited chatbot texts and in 0.5% of texts people typed, but also in 14% of published human writing. It counts as one sign among several.
What does "Probably not AI" mean?
It means the text shows few signs of AI writing. In our tests 43% of human texts got this result, and 5% of AI texts. It is a strong hint that a person wrote the text, not a guarantee.