How to Find and Remove Invisible Characters in Text (ChatGPT, Word, PDF)
· 5 min read · by the ProFigma team

You paste a paragraph into your website, a form or a Figma file, and something is off. A word will not wrap, a link breaks, a search finds nothing, or your code throws an error on a line that looks perfect. Very often the cause is a character you cannot see. This guide explains what invisible characters are, where they come from, the problems they cause and how to remove them in a few seconds.
What is an invisible character?
Every letter, space and line break in a text is a Unicode character with its own code. Most of them you can see. Some take up no space at all, and others look exactly like a normal space but are a different character. You cannot tell them apart by looking, but software can, and it treats them differently.
The ones you meet most often:
| Character | Code | What it does |
|---|---|---|
| Zero-width space | U+200B | Takes no space. Marks a place where a long word may break. |
| Zero-width joiner and non-joiner | U+200D, U+200C | Join or separate letters. Also used inside emoji such as 👨👩👧. |
| Byte order mark | U+FEFF | An invisible mark that often sits at the start of copied text or files. |
| Soft hyphen | U+00AD | Invisible until a word breaks at the end of a line. |
| No-break space | U+00A0 | Looks like a space, but keeps two words on the same line. |
| Narrow no-break space | U+202F | A thinner no-break space, common in AI and French text. |
| Direction marks | U+200E, U+202E | Change the direction of text. Can make text show in a different order than it is stored. |
| Tag characters | U+E0000 to U+E007F | Invisible copies of letters. They can hide a whole sentence inside normal-looking text. |
Where do they come from?
- AI chat tools: text from ChatGPT and similar tools sometimes contains no-break spaces, narrow no-break spaces or zero-width characters, along with curly quotes and long dashes. OpenAI has said these are not a watermark.
- Word processors: Word and Google Docs add no-break spaces, soft hyphens and curly quotes while you type.
- PDFs: copying from a PDF often brings soft hyphens, odd spaces and broken line breaks.
- Websites and apps: some sites add zero-width characters to help long words wrap, and messaging apps add their own marks.
- Hidden prompts: some people add tag characters on purpose, to hide instructions for AI tools inside a page or a document.
The problems they cause
- Layout: words that will not wrap, gaps that cannot be deleted, or text that breaks in the wrong place on a website or in Figma.
- Links and email addresses: a zero-width space inside a URL makes it fail, even though it looks correct.
- Code and data: a byte order mark at the start of a CSV or JSON file, or an invisible character in a variable name, causes errors that are very hard to find.
- Search and spreadsheets: a name with a no-break space will not match the same name typed normally, so search, filters and formulas miss it.
- SEO: hidden characters in page titles, descriptions or web addresses can make them look broken in Google.
How to find and remove them, step by step
The quickest way is our free Invisible Character Detector. It runs in your browser, so your text is not uploaded.
- Paste your text. Copy the text from ChatGPT, Word, a PDF or a website and paste it into the box. You can also press Try an example to see how it works.
- Press Check text. The detector reads every character and gives you a report: how many hidden characters, odd spaces and fancy punctuation marks it found, with the name and Unicode code of each one.
- See where they are. Your text is shown again with each hidden character marked in colour. Red marks are invisible characters, orange marks are odd spaces, purple marks change text direction and pink marks are hidden prompt text. Hover a mark to see what it is.
- Clean and copy. Press Remove hidden characters to delete the invisible ones and turn odd spaces into normal spaces. Press Clean everything if you also want plain straight quotes, simple hyphens and three dots instead of the ellipsis character. Then press Copy clean text.
The detector keeps emoji and the letters of every language as they are, so text in Spanish, Arabic, Japanese or any other language is safe to clean.
How to remove them by hand
If you prefer to work inside your own tools:
- Microsoft Word: press Ctrl + H, click More, then Special, and choose Nonbreaking Space or Optional Hyphen in the Find box. Leave Replace with as a normal space or empty, then click Replace All.
- Google Docs: paste the text into a plain text editor such as Notepad first, then copy it back. This removes most formatting, but not every invisible character.
- VS Code: invisible and unusual characters are highlighted with a yellow box. Use Find with regular expressions on and search for
[\u200B-\u200D\uFEFF\u00A0]to find the common ones. - Excel and Google Sheets: use
=SUBSTITUTE(A1, CHAR(160), " ")in Excel to replace no-break spaces with normal spaces. TRIM alone does not remove them.
Does removing invisible characters hide AI writing?
No. AI detectors judge the writing itself: word choice, sentence patterns and how predictable the text is. Invisible characters play little or no part. Removing them gives you clean text that works everywhere, but it does not change how the text reads or who wrote it. Hidden characters also do not prove that a text was written by AI, because they come from copy and paste, PDFs and word processors too.
Quick tips
- Check AI or Word text before it goes into WordPress, Webflow, a newsletter or a form.
- Use Remove hidden characters for normal writing, and Clean everything for code, CSV files and data.
- Line breaks are normal characters, not hidden ones. You only need to see them when the layout of the text matters.
- If a link or an email address fails and looks perfect, paste it into the detector first.
Ready to try it? Open the free Invisible Character Detector and paste your text.


