Skip to content

Text checks

ChatGPT Watermark Detector for Text

Paste a ChatGPT answer and Slop or Not's AI text detection model tells you how likely it is that AI wrote it. The result also lists any hidden and typographic characters that came along with the copy, and gives you a cleaned version. What it can't read is textGrain, the word-choice watermark OpenAI announced on 5 October 2026. Only OpenAI's detector reads it, and OpenAI is limiting access to approved researchers and expert organizations at first. Checking a ChatGPT image instead? Switch to the Photo tab. Text is checked on Numen's own servers, away from outside AI companies.

ChatGPT Watermark: What OpenAI's textGrain Means for Text

Catches 99.5% of AI text at a 5% false-positive rate, scored by RAID, the public benchmark, not by us.

Paste a paragraph or a whole thesis. The iPhone and Mac apps have no word cap and no monthly word budget.

English text only.Try an example0/20000

AI detection is probabilistic. Results can vary with new AI models and evasion methods. Human writing that sounds formal can be flagged as AI.

Scored by a fine-tuned transformer model, not a word-pattern checker.
  • Paste English textSchool essays, product reviews, articles, and long posts, from 150 words up to 20,000 characters. On a long text, the score covers about the first 18,000 characters.
  • ResultA probability verdict, AI score, and readability analysis.
  • PrivacyChecked on Numen's private Mac server and never passed to an outside AI company. For text that stays on your device, use the app.

What it checks

Scan for ChatGPT Traces, Then Read the Writing Itself

The narrow no-break space at U+202F, plus zero-width spaces and joiners, word joiners, and byte-order marks

Smart punctuation the chat window bakes in: curly quotes, em dashes, ellipses, and fullwidth symbols

A verdict on the writing itself from the RAID-scored text model, which needs no watermark at all

Good Uses

  • Checking whether a passage copied out of ChatGPT still carries chat-window artifacts
  • Reviewing a submitted essay, report, or pull request description before deciding what to ask its author
  • Cleaning a ChatGPT draft of stray Unicode before it goes into a CMS, a code comment, or a database

Limits to Know

  • The checker cannot read textGrain, OpenAI's word-choice watermark: reading it takes OpenAI's secret key, and OpenAI said access to its detector would initially be limited to approved researchers and expert organizations
  • A clean scan proves nothing: ChatGPT text pasted through a plain-text editor carries no traces, and human writing picks up the same characters from word processors
  • Cleanup strips the mechanical traces and nothing else; US and UK spelling conversion runs in the iPhone and Mac apps, not the online scan

How to check ChatGPT text for watermarks and hidden characters

  1. 1

    Copy the answer out of ChatGPT

    Select the answer in the chat window and copy it. Copying from the rendered chat is what carries the typographic characters across; an export to plain text usually drops them first.

  2. 2

    Paste it into the checker above

    Paste the answer in. It goes to Numen's server for the character scan, is processed in memory, and is discarded with the response.

  3. 3

    Run the check and read the traces

    Nothing appears until you run the check. Under the verdict, the fingerprints panel counts every trace it found and lists each narrow no-break space, zero-width character, smart quote, and lookalike letter by Unicode name with how many times it appears.

  4. 4

    Check the writing, then copy it clean

    Run the AI text check for a verdict on the words themselves. One click copies a version with the mechanical traces removed.

FAQ

ChatGPT Watermark Detector Questions

Does ChatGPT watermark its text?

OpenAI announced on 5 October 2026 that textGrain, a watermark that adds an invisible statistical signal to the model's word choices, would reach eligible ChatGPT and Codex text in the EU over the coming weeks. OpenAI also let API customers around the world opt in on select models from that day. The checker cannot read that signal; it reads the typographic characters the chat interface bakes into whatever you copy out.

What is U+202F, and why is it linked to ChatGPT?

U+202F is the narrow no-break space, an ordinary typographic character used before punctuation in French and inside some number formats. It picked up its AI reputation when readers noticed it turning up in ChatGPT output in 2025, often before a colon or a percent sign. Treat a hit as a hint worth following, not a verdict: any passage that has been through a typesetting tool can contain one.

Does OpenAI offer an official ChatGPT watermark checker?

Not to the public at launch. OpenAI opened applications for its text watermark detector on 5 October 2026, but said access would initially be limited to approved researchers and expert organizations. Its earlier AI text classifier was withdrawn in 2023 for low accuracy, and the public provider-side tooling covers images instead: C2PA Content Credentials on OpenAI image output, and SynthID on the supported OpenAI and Google image variants.

Do ChatGPT images carry a watermark?

Yes, and they are a different check. OpenAI image output can carry C2PA Content Credentials and a SynthID watermark, which Slop or Not reads directly rather than guessing at. Use the ChatGPT image detector for files; the check here is for pasted writing.

How is this different from the AI text detector?

The AI text detector asks one question about any writing: does it read as machine-generated? The check here starts narrower, with the characters ChatGPT specifically tends to leave behind, and then hands you the same text verdict for the writing itself. Character traces are fast and concrete but easy to lose; the text verdict survives a copy through a plain-text editor.

How do I remove a ChatGPT watermark from text?

Slop or Not's cleanup removes or normalizes hidden characters in one click: invisible Unicode goes, lookalike letters are swapped for Latin ones, smart punctuation turns plain, non-breaking spaces become ordinary spaces, and no word is rewritten. OpenAI's textGrain watermark, announced on 5 October 2026, lives in the word choices themselves, and no character cleaner, ours included, touches it. According to OpenAI, changing the words weakens detection.

Does removing the hidden characters make ChatGPT text undetectable?

No. The text model reads sentence structure and word choice, so stripping characters does not change what it sees. That model caught 99.5% of AI text at a 5% false-positive rate on RAID, a public benchmark we do not run, with 11 adversarial attacks in the test set. Character cleanup is for tidy text, not for hiding its origin.

What if the scan finds no hidden characters?

It means the text carries no mechanical traces, and nothing beyond that. ChatGPT output routed through a plain-text editor, an API call, or a code block often arrives clean, and plenty of human writing arrives with curly quotes and em dashes from a word processor. Run the AI text check when you want an answer about the writing rather than the characters.

How do I check ChatGPT text in Microsoft Word or Google Docs?

Select the passage, copy it, and paste it into the checker above, then paste the cleaned version back over the original. Word can show paragraph marks and tabs but not zero-width characters or lookalike letters, and Google Docs has no view for them at all. Both also add curly quotes and em dashes through autocorrect, so some of what the scan reports will be the word processor rather than ChatGPT.

Does my text leave my browser?

Yes. The character scan and the cleaned output are produced in memory on Numen's server and discarded with the response, so the scan stores nothing. The AI text check is a separate step and travels to Numen's private Mac server, away from third-party AI providers. The iPhone and Mac apps run both checks on-device.

Can I check ChatGPT text offline or from the command line?

Yes, on Apple devices. Copy an answer out of the ChatGPT app and the iPhone and Mac apps run both the character scan and the text check on-device, with no network involved. They also add spelling conversion between US and UK English that the online scan does not do. On the Mac, Pro ships a slop command and an MCP server, so piping the clipboard through slop cleanup handles it in one line, and Claude, Codex, or Cursor can call the same check.

Run the Next Check on Your iPhone or Mac

The Slop or Not app checks images and text on-device, so nothing you check ever leaves your device.

Download on the App Store