Sentence-level highlighting
The main figure is a good start, but the highlighting is really helpful. ChatZero highlights the sentences that appear most AI-like so you can easily check them.
ChatZero is a free AI detector for ChatGPT, GPT-6, Claude and Gemini. Paste your text into the AI percentage checker to get an AI score with highlighted sentences in seconds.
Many AI checkers provide you with just a percentage. We’ve made this tool a bit more helpful by giving you a general score and showing you the sentences that boosted or lowered your overall AI score.
The main figure is a good start, but the highlighting is really helpful. ChatZero highlights the sentences that appear most AI-like so you can easily check them.
Once you have checked your text you’ll be shown a score between 0-100, with a brief description such as "looks mostly human" or "mixed." This gives an approximate idea of how AI-like the text appears to be.
Most of the time people will be looking for a ChatGPT checker, this tool has been optimized for this so it works for newer GPT-6 style text. It also checks for patterns common to Claude, Gemini and other chat models. It won’t tell you which model was used to write any particular sentence.
You can check your text without having to sign-up or install anything. Just paste your text and click check, you’ll receive your results in your browser. We may implement optional logins in the future to save your history, the basic detection tool will always be free to use.
Your essays and drafts are personal so they should be treated as such. We will never publish any text you paste into ChatZero and it isn’t a showcase for students. See our privacy section below to see what we do and don’t store.
Most people will paste their text. If you have your draft in a document you can upload a .txt, .docx or .pdf file. You can then check the extracted text before checking. If the file can’t be read clearly you’ll receive an error.
Ready in a minute, not a homework session
This tool will be easy to get to grips with. All you need to do is paste some text into the box.
01
Paste your essay, article or email. Alternatively upload a file and have a look at the text. Paste at least 50 words. Short texts won’t give the tool much to work with.
02
Click ‘Check Text’. ChatZero will look at patterns in the text that are commonly found in AI generated text, such as repetitive sentence structures and generic language. It will also give an overall score.
03
Start by looking at the overall score. Then look at the highlighted sentences. Read through them to understand how the text has been scored. A text that has just one sentence that looks robotic will be scored differently than a whole page with repetitive language.
04
As a student this can help you find any parts of text that still look like they were written by a machine. As a teacher you should treat this as one piece of input, not the final word.
AI generated text will often have clear similarities with each sentence being very fluid and similar in length. ChatZero will highlight these issues and other common issues chatbots will have. For example bullet points, using common phrases such as "it's important to note," and having nice sentence endings. While this does not prove that it is AI generated, it shows which sentences appear to be the most fake.
Who this is for
The busy periods for this would typically be in school as it was designed to help with essays and in class writing.
However, if you have written the piece yourself it may be worth just running it through. If there are any parts flagged as looking like AI, try reading them out loud to see if they still sound like you. If you have used ChatGPT or any tools, make sure you are following the rules set by your school.
While this can help point out any abrupt changes in writing style, it doesn’t tell you how the essay was written. It may be worth using ChatZero on any drafts, in class writing and speaking to the student.
If you are commissioning work or have lots of editing to do it can be useful to run it through. If it scores highly you can ask a few questions.
Once again if you are applying for a job or submitting an article it can be a good way to check but shouldn’t be the only reason to accuse someone of cheating.
People often ask whether AI detectors are accurate. They’re a great tool but aren’t perfect. ChatZero can get things wrong both ways. It can fail to flag AI generated text that has been greatly rewritten and can flag human text that is very formal, very structured or written in a second language.
A moderate result (30-40%) does not necessarily mean the text is bad. Usually this means the text is a mix or is difficult to determine. This is quite common with school work. Short texts can also be unreliable so we ask you to write at least 50 words. Usually longer texts will give you a better understanding.
Another issue with this tool is formulaic texts (lab reports, legal text and outline templates). These can seem like they’ve been written by an AI. However, the tool cannot prove you didn’t use an AI. Websites cannot see what you have written in your draft. If you need to prove how a piece was written, using the version history in Google Docs, previous drafts and notes will be much more effective.
If you need this result for something important (a grade, job, disciplinary meeting etc.) do not take this page as the final say. Use it as a guide and look at the actual text.
Sometimes human texts can be flagged, especially if they are well written, repetitive or have an unusual tone.
Usually an average result will make you go back and look at the highlighted sentences.
ChatZero doesn’t decide if someone has cheated. Humans do and should use multiple pieces of information.
You may find these terms are often confused. An AI detector determines how likely text is to be generated by an AI model (like ChatGPT). Plagiarism detectors look for copied content. Original content can be AI generated and copied content can be human written. This tool detects the former but if you need to detect copied content, use a plagiarism tool.
You should not be worried that any essays you paste into this free tool will be published anywhere. We will never publish your text and do not sell it as part of a dataset of student essays.
The full text can be found in our Privacy Policy.
Common questions
ChatZero is a free AI detection tool. Just paste in some text or upload a file and the tool will work out how likely the text is to be AI generated (by models such as ChatGPT/GPT-6/Claude/Gemini). It will also highlight parts of text which are likely to be AI generated.
Paste in your text into the box at the top or upload a .txt/.docx/.pdf file. Once you have put in at least 50 words hit 'check'. You will then get a score for how likely it is to be AI generated and highlighted sentences. No app is required.
No. You do not need to sign up, put in an email or pay. Although there may be some optional features added in the future (such as saving your history), the main feature will still be available.
You can upload .txt/.docx/.pdf files. You will then see the extracted text in the box. If it can’t be read you will get an error.
AI detectors look for patterns in the text that are much more common in AI generated texts. This can include repeating vocabulary, very smooth transitions between sentences and very generic wording. ChatZero uses these patterns to give an overall score and a score per sentence. It cannot see how the file was created.
No. While the tool mainly focuses on detecting ChatGPT (as most people will be thinking of this), it looks for patterns common to many popular chat models (including Claude and Gemini). It will not tell you which model was used to create the sentences.
ChatZero is mainly for English. Results on other languages may be unreliable, so we do not treat those scores as meaningful yet.
We can’t say how accurate we are honestly, and you shouldn’t make any serious academic or hiring decisions based on this. Try looking at the highlighted sentences and the overall context.
Like all AI detectors ChatZero isn’t perfect. It can struggle with heavily edited AI writing and also pick up on formal, repetitive or very consistent human writing.
It can be quite common for the result to come back as medium, meaning it’s a bit of a mix or the tool simply can’t decide. This is very common with school work, as some paragraphs may have been heavily edited. Take a look at what sentences have been highlighted.
This can happen more than you’d expect. Text can be flagged as seeming like AI due to it being very concise in an academic sense, using templates or being written in a second language. If you wrote it yourself then it’s best to look at the flagged lines and if you need to explain your work then showing drafts or version history will be a much better option.
We can’t prove a negative. ChatZero will only score the text you have pasted, we can’t see how you typed it. Showing previous drafts, outlines and version history will be a much better option to show your process. If a teacher or editor has raised concerns then it’s usually best to discuss your process with them.
Yes! Students use it to quickly check if their essays are AI and teachers use it as an extra step to check if a paper feels off. This should be the first step in taking a closer look at the text. Policies around AI in the classroom will come from the school.
No. ChatZero is an estimate of whether text seems like it was generated by AI. Plagiarism checks are looking for text that has been copied from elsewhere. These are two separate issues and a paper can fail one and pass the other. If you need a plagiarism check use a tool designed for this.
We won’t publish any text you paste and we won’t use student essays as examples. Our full policy can be found in our Privacy Policy. This includes any vendors we may use to help run our detection tool.
It is completely free to use, works in your browser and there is no need to sign up. Simply paste your text at the top of the page and receive your score and highlighted sentences in a few seconds.
Check your text now