AI watermark remover
Paste text from ChatGPT, Claude, Gemini or anywhere else to reveal invisible watermark characters, then clean them out in one click.
Preview
- Hidden
- Em dashes
- 0
Purple diamonds are invisible characters. Amber dashes are em dashes, which the cleaner replaces with commas.
How to use the AI watermark remover
- 1 Paste your text and click Scan for hidden characters.
- 2 Review the highlighted characters and remove them one by one, or all at once.
- 3 Copy the clean text.
What an AI watermark in text is
People use the term for two different things. The first is invisible Unicode characters: zero-width spaces, narrow no-break spaces and similar marks that sit in the text without showing on screen. This tool finds and removes those. The second is a statistical watermark, such as Google DeepMind’s SynthID Text, which is a pattern in the choice of words. There is no character to delete, so no character cleaner can remove it.
Are the hidden characters deliberate?
Often not. Invisible characters also slip in when you copy from web pages, PDFs and word processors, and some come from how an assistant formats its output. Nobody publishes a full list of which product inserts what, so treat any character you find as a formatting artefact worth cleaning, and not as proof of where the text came from.
What it removes
Zero-width spaces (U+200B), non-joiners and joiners (U+200C, U+200D), left-to-right and right-to-left marks (U+200E, U+200F), narrow no-break spaces (U+202F), soft hyphens (U+00AD), the Mongolian vowel separator (U+180E), word joiners and invisible operators (U+2060 to U+2064) and byte order marks (U+FEFF). Narrow no-break spaces become normal spaces. Em dashes are highlighted separately and replaced with commas.
| Character | Code | What cleaning does |
|---|---|---|
| Zero-width space | U+200B | Removed |
| Zero-width joiner and non-joiner | U+200D, U+200C | Removed |
| Narrow no-break space | U+202F | Becomes a normal space |
| Soft hyphen | U+00AD | Removed |
| Directional marks | U+200E, U+200F | Removed |
| Byte order mark | U+FEFF | Removed |
| Em dash | U+2014 | Becomes a comma |
When cleaning changes your text
Zero-width joiners and directional marks do real work in some languages and in emoji, including Arabic, Hebrew, Persian and several South Asian scripts. If your text uses those, check the preview and remove characters one at a time instead of cleaning everything.
Use it responsibly
Cleaning text is not a licence to pass off AI-written work as your own. Follow your school’s or employer’s rules on AI use, and disclose it where required. Quizard is built for learning, so we would rather you practise from your own material.
AI watermark remover FAQ
Does ChatGPT put watermarks in text?
Invisible characters such as narrow no-break spaces have been reported in text copied from ChatGPT and other assistants, but they are not a reliable signature, and they can also come from other software. Paste your text here to see what is actually in it.
How do I remove an AI watermark from text?
If the watermark is made of invisible characters, paste the text here, click Scan for hidden characters, then Clean text. Your words stay as they are. A statistical watermark such as SynthID Text is a pattern in the wording, so it cannot be removed this way.
How do I clean AI text and remove AI markers?
Paste the text into the tool. It lists invisible characters such as zero-width spaces and other markers, shows where each one sits, and removes them in one click, leaving your words untouched. Read the result before you use it: cleaning formatting artefacts does not change what the text says, and it is not a way to pass off AI writing as your own.
Will removing hidden characters make AI text undetectable?
No. AI detectors mostly judge writing patterns and predictability, not invisible characters, so this tool does not change how a detector reads your text. Detectors are also imperfect and can flag human writing, so a score is not proof either way.
What specific characters does this tool remove?
Zero-width spaces (U+200B), zero-width non-joiners (U+200C), zero-width joiners (U+200D), left-to-right and right-to-left marks (U+200E, U+200F), narrow no-break spaces (U+202F), soft hyphens (U+00AD), the Mongolian vowel separator (U+180E), word joiners (U+2060), invisible operators (U+2061 to U+2064) and zero-width no-break spaces (U+FEFF). Em dashes (U+2014) are replaced with commas.
How do I know if my text contains hidden characters?
You usually cannot see them. Signs include odd spacing, a cursor that seems to jump over nothing, and searches that fail on words you can see. The easiest check is to paste the text here and click Scan for hidden characters. Each one is highlighted and named.
Is my text uploaded?
No. The scan and the cleaning both run in your browser, so your text never reaches our servers.
Is it ethical to use this for academic or professional work?
Removing stray formatting characters is ordinary text cleaning. Submitting AI-written work as your own is a separate question that your institution’s rules answer, so check them and disclose AI use where required.
Your whole semester, in one study workspace.
Quizard turns your own lecture slides into questions and flashcards that cite the slide, and shows what you have not covered.