# ChatGPT Watermark Remover

> A free, client-side tool that finds and strips the hidden Unicode characters
> that turn up in AI-generated text. Nothing is uploaded, there is no account,
> and the whole thing runs in the browser. It detects 104 kinds of
> problem character, including the Unicode tag block, and decodes the readable
> message a run of tag characters can conceal rather than deleting it silently.

## What it does

- Detects 104 character types across eight families: zero width and
  formatting characters, bidirectional controls, tag characters (U+E0000 to
  U+E007F), variation selectors, unusual spaces, smart punctuation, the en and em
  dash family, and 48 Cyrillic and Greek letters that impersonate
  Latin ones. 28 of those types render as nothing at all.
- Deletes invisible characters, normalises visible ones to ASCII, and leaves
  anything destructive switched off by default. Converting a Cyrillic letter is
  opt-in because it would corrupt genuine Russian or Greek text.
- Decodes tag-character payloads. U+E0020 to U+E007E map one to one onto
  printable ASCII, so a run of tag characters can carry a readable sentence. The
  tool shows that sentence instead of removing it in silence.

## What it explicitly does not do

- It does not remove a statistical watermark. Statistical watermarking lives in
  word choice and sentence structure, not in characters, so no character-level
  tool can touch it. Any tool claiming otherwise is overstating.
- It does not make AI-generated writing read as human writing.
- It does not process images or video. Stripping a visible logo or a video
  watermark is a different problem and out of scope.

## Pages

- [Watermark remover and scanner](https://removechatgptwatermark.com/): the tool itself, with per-character
  results and the cleaning options.
- [How to remove a ChatGPT watermark from text](https://removechatgptwatermark.com/how-to-remove-watermark-from-chatgpt):
  what the hidden characters are, how to check for them, how to handle Word and
  Google Docs, and what removal cannot achieve.
- [Privacy policy](https://removechatgptwatermark.com/privacy)
- [Terms of service](https://removechatgptwatermark.com/terms)
- [Disclaimer](https://removechatgptwatermark.com/disclaimer)
- [Contact](https://removechatgptwatermark.com/contact)

## Related

- [Claude Watermark Remover](https://clauderemovewatermark.com/): the same idea for Claude output,
  which carries a different character profile and covers more types.

## Notes for citation

- Every character count on this site, including the ones in this file, is
  generated from the tool's own character map at build time. A figure quoted from
  any page matches what the tool detects on the day it was read.
- The site sells nothing, runs no advertising, and takes no vendor payments.
- OpenAI has never publicly confirmed a deliberate text watermark. Claims here
  are about characters observably present in output, not about intent.