How to check text for hidden characters

Paste the text into a checker that shows you the actual codepoints, then copy the cleaned version back out. It takes about a minute, and nothing has to leave your machine to do it.

The steps

1. Paste the text into the AI text detector. Findings appear as you type, and the scan runs in your browser — open your network tab and watch if you want to confirm nothing is sent.
2. Read what it lists. It checks 539 codepoints across seven groups: zero-width characters, unusual spaces, bidirectional controls, Unicode tag characters, variation selectors, control characters, and letters swapped for look-alikes from another alphabet.
3. Copy the cleaned version. Invisible characters are deleted and odd spaces are normalised to ordinary ones, so the text reads identically but imports cleanly.

One thing to check before you clean: if the text is in Arabic, Persian or Hindi, or contains multi-part emoji, some of those zero-width characters are load-bearing. Remove them and the text renders wrong.

What the result tells you

Characters found. The text passed through software that inserts them — a CMS, Word, a PDF export, a translation tool, a browser copy, or a model. Which one, the character itself cannot tell you.

Nothing found. Then nothing. Most writing contains no invisible characters, so an empty result is not evidence a person wrote it.

Either way it is not authorship evidence. Hidden characters are not an AI watermark, and the marks that are real sit in word choice rather than in stray Unicode, so no amount of character scanning will turn one up.

What is not a hidden character

Em dashes, curly quotes and ellipses get called AI tells constantly. They are visible punctuation, word processors have inserted them for decades, and a checker that reports them as findings is flagging ordinary writing. Cleaning is worth doing because invisible characters break imports, corrupt e-book formatting and confuse screen readers — reasons that hold whoever wrote the text.