Everything runs in your browser. Your text is never uploaded.
List every character
| Line:Col | Code | Name | Note |
|---|
Everything runs in your browser. Your text is never uploaded.
| Line:Col | Code | Name | Note |
|---|
Paste any text and this invisible character detector shows every character you can’t see: zero-width spaces, word joiners, byte-order marks, direction controls, unusual spaces, soft hyphens and control characters. Each one is marked in place with a short label such as ZWSP or NBSP, so you can see exactly where it sits. Then it gives you a clean copy with one click.
It also catches two tricks most cleaners miss:
а looks exactly like a Latin a, which is how fake links like pаypal.com are made. Words that mix scripts are highlighted, and you can swap the look-alikes back.Some invisible characters do a job. The zero-width joiner holds family and skin-tone emoji together, variation selectors make ❤️ show in colour, and Persian, Hindi and other scripts need joiners to shape letters correctly. Blindly deleting them breaks emoji into pieces and misspells words. With Keep joiners that emoji and scripts need ticked (the default), the tool keeps these and marks them in grey instead of red.
You may have read that ChatGPT hides invisible characters in its answers as a watermark. As far as is publicly known, it doesn’t do this on purpose. In April 2025 some users found narrow no-break spaces in answers from newer OpenAI models; OpenAI described it as a quirk, not a watermark. Real AI watermarks, such as Google’s SynthID for text, work by nudging word choices, which no character tool can see or remove. What this tool does reliably is find and remove any hidden characters that are actually in your text, whatever put them there.
Paste the text into the invisible character detector above. Every hidden character is marked in the text with a label, counted by type and listed with its line and column. If none are found, the tool says so.
U+200B is a character with no width at all. It was meant to mark where long words may break, but in practice it mostly arrives by accident when text is copied between apps. It breaks searches, word counts, passwords and code because two strings that look identical no longer match.
No. AI detectors judge wording and sentence patterns, not invisible characters. Cleaning fixes the text itself; it doesn’t change how it reads.
A zero-width space, a no-break space or a byte-order mark copied from a web page or chat window is a common cause. Paste the code here, and the marked view shows the culprit’s exact line and column.
It is hiding text in invisible Unicode tag characters. People see nothing, but many AI models read the hidden text and may follow it as an instruction. If this tool shows a hidden message, remove it before pasting the text into an AI assistant.
No. The invisible character detector runs entirely in your browser, so you can safely check private documents.
The invisible character detector is free, needs no sign-up and works on phones too. Background on how Unicode defines these characters is on the Unicode Consortium website.
Want the full story? Read our guide Does ChatGPT Add Hidden Watermarks? How to Find and Remove Invisible Characters. To also strip markdown symbols and asterisks from AI answers, use the ChatGPT Text Cleaner.
Short, practical reads that pair with this tool.
Free, in your browser, no sign-up.