Free Claude Watermark Remover
This page does not remove Claude's own watermark: Anthropic says it is a statistical mark applied while the model picks its words, with no hidden characters, so no character cleaner can remove it. What it does is show and remove the invisible characters and unusual spaces that come along in a paste, right in your browser: no upload, no account, no limit, and your visible words come back as they were.
Last updated: September 2026
Cleaning happens in your browser: that text never leaves this page, and there is no upload, no account and no limit. Rewrite is the exception: that one sends your text to our server.
What Anthropic has said about Claude's watermark
Anthropic's help centre says Claude models launched on or after 2 August 2026 support machine-readable marking from launch, with earlier models being added during a transition, as part of the EU AI Act's Article 50 transparency code. The mark is described as imperceptible and "woven directly into the text", and it applies to Claude's apps, the API, Claude Code and Anthropic's other products worldwide. Files Claude creates get Content Credentials (C2PA) metadata instead; for an image file such as a PNG, JPG or SVG, that is information stored in the file, not something painted into the picture. Anthropic's own article on how Claude's text watermarking works (14 August 2026) describes a version of the SynthID-Text approach: it changes which word the model picks, nothing is added to the text and there are no hidden characters. It says light editing probably won't remove the mark completely and a complete rewrite where every word is replaced (paraphrasing) will, that it is sparser in factual passages and code, and that detection works only through an API in private preview, giving a likelihood rather than proof.
So we do not know whether the mark involves any character at all, and we will not pretend this cleaner removes it. If it works the way Google's SynthID text watermark does, by steering which words get picked, nothing that edits characters can touch it; changing it would mean rewriting the text.
What a Claude paste does contain
What you can actually see in a paste from Claude is the model's typography, and it is well documented: several bug reports on Anthropic's own claude-code issue tracker ask the model to stop writing em dashes, en dashes, curly quotes and the single-character ellipsis into code and commit messages. Those arrive in any paste, from the chat, the desktop app or Claude Code. The copy button under a reply, and the Copy button on an Artifact, give you markdown source (## headings, **bold**, code fences), which are visible characters this tool leaves alone. Odd spaces and a hidden character such as a zero-width space are not something Claude is known for; when they appear, they usually came from a web page, a PDF or a document you pasted into the conversation first.
What the cleaner did with a Claude-style paste
We ran this page's cleaner on one sentence written with the typography the bug reports describe: two curly apostrophes, an em dash, an en dash in a year range, curly double quotes and an ellipsis character.
| Found by the cleaner | Count | After cleaning |
|---|---|---|
| Right Single Quote (U+2019) | 2 | straight apostrophes, with the quotes-and-dashes option on |
| Em Dash (U+2014) and En Dash (U+2013) | 1 + 1 | plain hyphens, same option |
| Curly double quotes | 2 | straight quotes, same option |
| Ellipsis (U+2026) | not counted | left as it was: the cleaner has no rule for it |
| Invisible characters or odd spaces | 0 | nothing to remove |
With the option off, the cleaner found nothing at all in this sentence, which is the typical Claude result: the cleanup is about typography, not hidden characters. This was a test string we wrote to match that pattern, not a captured reply; the counts are what this page's own cleaner reported when we ran it on 2 October 2026.
Using this as a Claude watermark remover
People looking for a Claude watermark remover, or an AI watermark remover for text in general, usually need one of two things. If a pasted Claude reply breaks a form, a CMS field or a script, this page finds the cause: it is a watermark remover free of daily caps and a watermark remover online that stays inside the browser tab. If the goal is to get past Anthropic's mark, no cleaner does that (statistical watermarks are not made of characters), and the Rewrite button is a writing aid, not a way around detection. The cleaner is the same one linked from every assistant's page on this site; the explanation around it is what differs.
but the ordinary rendering debris still differs depending on which button you pressed.Claude Code, the API and the desktop app
A terminal does not add rendering characters the way a web page can, so text out of Claude Code or straight from the API console is close to what the model generated, and the dashes and quotes the model chose are still there. In the desktop app, in Claude in Slack and in a shared conversation link the reply is rendered first, so a select-and-copy gives you plain text while the copy button gives you markdown. Either way the cleaner scans the characters that arrived, not the app that sent them.
All assistants, one tool
The cleaner is the same on every page: it matches characters, not sources. See the full list of what it removes.
Questions
Does this remove Claude's watermark from my text?
No. Anthropic says its mark is a statistical watermark in the choice of words, not in characters, so no character cleaner reaches it, and this tool makes no claim to. It removes invisible characters and unusual spaces, and optionally straightens quotes and dashes, without changing your words. There is a separate Rewrite button for people who want their wording changed; that is a writing aid, not a way around detection.
Is my text uploaded anywhere?
The cleaner runs entirely in your browser: that text never leaves the page, there is no account, and there is no limit on how much you clean. The separate Rewrite button is the one exception: it sends your text to our server, which passes it to Anthropic, whose AI model rewrites it. Only Anthropic receives it, to rewrite it, and Anthropic does not train its models on it. We keep no copy of your text or the rewrite.
Why does pasted Claude text break my code or CMS?
Usually because of the typography: a curly quote where a straight one was expected breaks a string literal in a code editor, and an em dash in a command-line flag is not a hyphen. Invisible characters from a document you pasted into the chat earlier do the rest; they survive a copy-paste but are not what they look like.
Is there a watermark remover for videos too?
Not on this page: this one works on text only. For a video's on-screen watermark, WatermarkNinja is the tool in this family built for that.
Is there a Claude watermark remover skill, or one on GitHub?
Not from us: TextNinja is not a Claude Skill, has no GitHub repo and is not connected to any similarly named project. It runs as a free web page: paste your text and get the same invisible-character cleanup, nothing to install.
Is there any AI behind the cleaner?
No. The cleaner is plain character matching against the Unicode standard, in your browser. Only the separate Rewrite button uses a language model, run by Anthropic on our behalf.
Does an Artifact copy differently from a chat reply?
Mostly in format: an Artifact's Copy button gives you the raw code or markdown, while selecting text in the chat gives you the rendered version as plain text. The model's own dashes and quotes are in both. This tool checks either the same way and shows you exactly what it found.
What this tool cannot remove
It removes named invisible characters (Unicode ), turns space variants into an ordinary space and, if you leave the option on, makes typographic replacements. It also clears the three private-use ranges, including the tag range U+E0000: U+E007F, and it treats text from a PDF, Word or a web page exactly like text from an assistant. It cannot touch statistical watermarks such as Google's SynthID, which live in the choice of words, not in any character, so no character cleaner can. Anthropic marks the text of its newer Claude models with a statistical watermark in the choice of words and says nothing is added to the text (anthropic.com), so no character cleanup reaches it; assume it does not. Sources we checked: Anthropic at anthropic.com and support.claude.com, and OpenAI's content-provenance page.
By Sam Ridder, I build and run TextNinja on my own hardware, on my own. Who I am.