Free watermark check

Invisible character detector

Every hidden Unicode character in your text, named, classified and located — plus a clean copy you can actually submit.

Paste text

Paste, do not retype — invisible characters only survive a copy.

Upload PDF / DOCX / TXT

Click to choose a file

No account, no word limit

We never ask for a name, an email address or a university.

No invented number

Separate findings, not one percentage. Style is never counted as evidence.

Says what it cannot do

Claude's statistical watermark is checkable by nobody today, including us.

The short answer

This free detector reveals invisible Unicode characters in any text: zero-width spaces (U+200B), zero-width non-joiners and joiners, the byte order mark, Unicode tag characters (U+E0000–U+E007F), variation selectors, bidirectional controls, soft hyphens and unusual spaces. Each finding is shown with its code point, its position in your text and what it typically means. It also detects words mixing Latin with Cyrillic or Greek lookalike letters, and decodes hidden messages carried by zero-width characters. You can also copy out a cleaned version. Free, with no account and no word limit.

Exact positionsDecodes hidden messagesClean copy in one clickNo account

Characters this detector finds

  • Zero-width: U+200B, U+200C, U+200D, U+FEFF, U+2060 — invisible and almost never legitimate in prose
  • Unicode tag characters U+E0000–U+E007F — completely invisible, able to carry a whole ASCII message
  • Variation selectors U+FE00–U+FE0F and U+E0100–U+E01EF — one hidden byte each
  • Bidirectional controls — can make displayed text differ from stored text
  • Soft hyphens and unusual spaces — usually harmless, but they break search and parsing
  • Homoglyphs: Cyrillic or Greek letters inside otherwise Latin words

What finding them does not mean

  • It does not mean AI wrote your text. Websites, chat interfaces and word processors all introduce these.
  • It does not mean a Claude watermark is present — that is statistical, not character-based
  • A clean result does not mean text is human-written

How to find invisible characters in text

  1. 1

    Paste your text, or drop in the file

    Copy rather than retype — invisible characters only survive a copy-paste. PDF, DOCX and TXT are read locally, and a DOCX also gives up its editing time and save count.

  2. 2

    No account, no word limit

    You are never asked for a name, an email address or a university. The text is analysed in your browser and the analysed text is then stored on our EU servers for twelve months, never linked to your identity.

  3. 3

    Read each finding separately

    You get one card per layer, each with its own confidence, and no combined percentage. Character findings are definite. Style habits are not evidence and are labelled as such.

Why invisible characters matter in a thesis

They break things quietly, and the breakage surfaces at the worst moment. A zero-width space inside a citation key makes a bibliography entry vanish from the compiled document. A non-breaking space in a numeric field breaks a data import. Soft hyphens defeat text search, so the reader who searches your PDF for a term finds nothing. Screen readers stumble over bidirectional controls. LaTeX often refuses to compile at all.

None of these announce themselves. The document looks perfect on screen, and the failure appears in the artefact you hand in.

Where do they come from?

Copying from a web page is by far the most common source: HTML rendering, emoji sequences and layout tricks all leave residue. Word processors add soft hyphens through automatic hyphenation and non-breaking spaces through autocorrect. PDF extraction introduces its own artefacts. Chat interfaces of every kind — AI assistants included — pass along whatever their rendering layer produced.

Deliberate use exists too, and that is what the mixed-alphabet check is for. A Cyrillic 'о' inside a Latin word does not arrive by accident from any of the above; someone put it there.

Is it safe to remove them?

For invisible control characters, yes — they carry no meaning in ordinary prose and removing them is what a careful copy-editor does. Our clean copy deletes them, converts unusual spaces to ordinary ones, and repairs lookalike letters only inside words that genuinely mix alphabets.

That last restriction matters more than it sounds. A blanket homoglyph replacement would silently corrupt a physics thesis, where Greek capitals are ordinary symbols, or any passage quoted in Russian or Greek. Inside a word that is already half Latin there is no such ambiguity, so that is the only place we touch.

We leave typographic punctuation alone. Em dashes and curly quotes are correct typography and belong in your document.

Frequently asked questions

U+200B, a Unicode character with no width and no visible glyph. It was designed to mark line-break opportunities in scripts without spaces. Because it is invisible but preserved through copy-paste, it is also the most common carrier for hidden data, and it silently breaks text search, LaTeX compilation and bibliography parsing.
Paste the text into this page. Every invisible character is listed by name and code point, with the exact positions where it appears in your text, and a note on what it usually means. The check is free and needs no account.
No. The most common sources are copying from a web page, word processor autocorrect and PDF extraction. AI chat interfaces can pass them along too, but so can any website. Invisible characters tell you about a text's journey between applications, not about who wrote it.
A character from one alphabet that looks identical to one from another — Cyrillic 'о' and Latin 'o', or Greek capital eta and Latin 'H'. Inside an otherwise Latin word, a homoglyph has no innocent explanation: it does not appear through typing, and it is used to evade text matching and to carry hidden marks.
Yes. After the check you can copy a cleaned version with invisible control characters deleted, unusual spaces normalised and lookalike letters repaired inside mixed-alphabet words only. Typographic punctuation is preserved, since it is correct and belongs in your document.

Sources

Related checks

Claude watermarkChatGPT watermarkFree watermark checkerAI check

The real question: will your work be flagged as AI?

That is what universities actually run, and those detectors are wrong about human writing more often than they admit. Our AI check publishes its false-positive rate beside every verdict. Free for the first 1,500 words.

Run the free AI check