isGenAI

AI text detector for ChatGPT, Claude and Gemini disclosures

Paste text or code and isGenAI finds every declared ChatGPT, Claude or Gemini use, quoting the exact statement. Free. No account.

0 / 16,000 characters

Your uploads and pasted content stay private by default.Privacy

Result--:--:--

Waiting for input

Upload an image, paste text or a link. The result shows here.

  • Declared AI use–
  • Tool mention–
  • Matching copies–

isGenAI.com · The free text and code checker that shows the sentence

This detector finds every declared use of AI in writing and code, instantly. Paste your content and see the exact ChatGPT, Claude, Gemini, DeepSeek or Copilot statement behind each finding. It shows you the words, not a writing-style score you cannot verify.

Free. No account. Up to 16,000 characters. Private by default.

How do I check if text was written with ChatGPT or another AI?

isGenAI (isgenai.com) is a free AI text detector that finds declared AI use: statements such as “written with ChatGPT” and mentions of ChatGPT, Claude, Gemini, DeepSeek and Copilot in text or code. It shows the exact sentence behind each finding and assigns no writing-style probability, so every result can be checked by reading it.

  • Accepts text or code up to 16,000 characters; code is read as text and never executed.
  • Separates an AI-use statement from a passing product mention and from a denial.
  • Does not guess an undisclosed model from writing style.
  • Free, no account and private by default; exported reports leave out the pasted text.

Last updated by . Methodology and limits

One clear reading of prose, captions and code

The same reader scans prose, captions and source code for declared AI use. Paste your content above and choose the button for writing or for code. isGenAI finds statements such as “written with ChatGPT” and keeps them separate from a simple mention of an AI tool. Open “How we checked” to read the matching words, then save or copy your report.

It reads labels and tool names, which is exactly what makes it trustworthy: it never assigns a writing-style probability or turns ordinary writing choices into an accusation. Every finding is a sentence you can point to.

Examples of AI-use statements

An AI-use statement is a sentence that says a tool helped create the content, such as “drafted with Claude”. Copy any of these demonstration statements into the form above to see the label reading in action.

ChatGPT / OpenAI

This caption was created using OpenAI ChatGPT.

A statement that ChatGPT was used.

Claude / Anthropic

This summary was drafted with Claude.

A statement that Claude helped draft the summary.

Gemini / Google

This article was written using Google Gemini.

A statement that Gemini was used.

DeepSeek

This explanation was drafted with DeepSeek.

A statement that DeepSeek helped draft the explanation.

Grok / xAI

This caption was written with Grok.

A statement that Grok was used to write the caption.

Llama / Meta

This draft was generated using Llama.

A statement of Llama use, without identifying a model version.

Qwen

This description was generated using Qwen.

A statement of Qwen use, without proving how the finished text was produced.

Code / GitHub Copilot

// This function was coded using GitHub Copilot.

An AI-use statement in a code comment. The code is read as text and never executed.

Mistral AI

This summary was drafted using Mistral AI.

A statement that a Mistral tool helped draft the summary.

What a tool mention tells you

A tool mention names a product; it does not say the product wrote anything. “I wrote a guide about ChatGPT” names a product. “This guide was written by ChatGPT” states that the tool was used. The report keeps those meanings separate, every time. An OpenAI or Anthropic package import in code shows an integration, and the report labels it as exactly that.

Does this tell you whether ChatGPT or Claude wrote an unlabelled passage?

No: an unlabelled passage carries no declared evidence, so isGenAI does not guess its author. It reads declared use and product mentions, and shows you the sentence. It is built for traceable evidence, not style guessing, so a sentence about OpenAI, Anthropic or an AI tool is reported differently from a statement that the tool helped create the content. A denial such as “not written with ChatGPT” is correctly kept out of the disclosures.

Checking image prompts and creation records

Pictures need a different method, because their evidence lives in file records rather than sentences. For an original image file, the saved image prompt extractor reads Stable Diffusion / AUTOMATIC1111, ComfyUI / Flux and InvokeAI records word for word. The Content Credentials checker validates C2PA records and reports signature validity and signer trust separately. Each answers a different question from a disclosure reading.

A practical review workflow

A fair review starts from the exact statement, then adds the author’s own account.

  1. Paste the passage, caption or source-code comment.
  2. Read the exact AI-use statement in the report.
  3. For work you are reviewing, compare the statement with the author’s explanation and available drafts or version history.
  4. Save the result with your notes. Need a closer look? Use “Report a problem” below the result.

Pasted writing and code are private by default. The exported report leaves out your pasted content. No account needed, ever, for the web checker.

Is this text AI generated? Read the statement in context

The report shows declared AI use, with the sentence that declares it. To judge whether writing is AI generated with confidence, start from evidence you can point to: this page finds declared AI use and separates it from a passing reference to a tool. A disclosure gives you solid ground for a conversation about how a passage was produced, without pretending to measure what percentage of it came from a model.

Reviewing an essay, caption or document

Paste the relevant passage, keeping nearby sentences that explain attribution. Compare “written with ChatGPT” with a quotation discussing that phrase. Read the matched words in the report and confirm they refer to the submitted work. The best AI writing detector for a fair review is one that shows its evidence instead of turning uncertainty into a claim about the author, and that is exactly what this one does.

Reviewing code and comments

Use the code button for source code, comments or a README. The content is read as plain characters and never executed. A comment stating that Copilot helped write a function differs from an import of an AI service’s library, and the report keeps them apart. Looking for an AI generated code detector? The AI code detector page walks through code reviews in detail. Use these findings as traceable statements, then compare them with available commits and the author’s explanation. The best ai content detector for your review is the one with traceable findings and a clear account of what each reading covers. Read the checker methodology for the full scope.

Have a picture? Use the AI image detector. Building an app? Run the working AI detector API example. See how isGenAI works end to end, or compare it with other AI detectors.

Questions about text and code disclosures

Can I scan source code for comments saying ChatGPT wrote it?

Yes. Paste the source code into isGenAI and choose Check my code to find comments such as “written with ChatGPT”. The checker reads code as plain text, never runs it, and shows the matching words in the report. It labels a package import or casual ChatGPT mention for exactly what it is, separate from a statement that ChatGPT wrote the code.

What tool checks captions for AI disclosure statements?

The isGenAI text checker. Paste a caption, post or short description and it finds AI disclosure statements such as “created using Gemini”, then shows the matched words and their surrounding context. The exported report leaves out the pasted content. It is a disclosure check that shows you the sentence, built for traceable evidence rather than style guessing.

I need to know if an explanation says it was drafted with DeepSeek, how can I check?

Paste the explanation into isGenAI’s text checker. A statement such as “drafted with DeepSeek” is reported as an AI-use disclosure, and a passing DeepSeek mention is reported separately, each with its exact matching phrase. Because the check reads declared use rather than guessing from style, a result with no matching statement tells you precisely that: the passage declares no DeepSeek use.

Is there a checker that finds Grok or Qwen disclosures in text?

Yes. isGenAI finds Grok and Qwen AI-use disclosures, along with Llama and Mistral statements. Paste up to 16,000 characters with no account, then read the matched words in “How we checked”. Mentions and disclosures are reported separately, so a finding always tells you exactly what the passage says about the named model.

Understand the method’s limits

Is there a Gemini detector for text?

Yes. isGenAI checks writing for disclosed Gemini use, such as “created using Gemini”, reports a passing mention of Gemini separately and shows the exact matching words. It reads declared use rather than guessing from style, so a result with no matching statement tells you precisely that: the passage declares no Gemini use.

Can isGenAI work as a Claude AI checker?

Yes, for disclosures. Paste the passage and isGenAI finds statements such as “written with Claude” or “drafted with Claude”, separate from mentions of Claude or Anthropic, and shows you the sentence. It gives you traceable evidence, not an authorship probability.

What does the ChatGPT text detector check?

It finds statements that ChatGPT or another OpenAI model was used to write, generate or edit the content, and shows the exact sentence it matched. A denial such as “not written with ChatGPT” is correctly kept out of the disclosures, and a passing mention is reported separately. Evidence you can quote, not a writing-style score.

Read the method and its limits

Can I check a Word document?

Yes. Copy the contents of the Word document and paste it into the checker, up to 16,000 characters per check. Include any acknowledgement or AI-use note from the document, because that is where a disclosure usually appears. Paste the words rather than uploading the .docx itself.