Not sure whether a piece of text is watermarked? Paste it below and this detector scans every character, reports whether hidden watermark characters are present, and shows you exactly what and where they are. Runs entirely in your browser — nothing is uploaded.
The scanner checks your text against every common family of invisible and format characters used to watermark or fingerprint text:
Zero-width spaces and joiners (U+200B, U+200C, U+200D, U+2060) and the byte order mark — the classic invisible markers slipped between letters.
Direction-control codes (U+202A–U+202E and the isolates) that can hide payloads or reorder how text is read without changing what you see.
Selectors (U+FE00–U+FE0F) and tag characters (U+E0000–U+E007F) — a known channel for smuggling machine-readable data inside plain text.
No-break, en, em and ideographic spaces plus soft hyphens that look ordinary but are not the standard space character.
Detecting a text watermark means finding invisible characters that a reader cannot see. Because these characters render with no width, you cannot spot them by eye — you need a scanner that reads the raw code points. Paste your text into the detector above and click Analyze Text: it inspects every character, counts any invisible ones, groups them by type, and marks their exact position so you can confirm what is inside.
This is useful whenever you receive text and want to know if it has been fingerprinted — a document, an email, a message, or code. The detector answers a simple question honestly: are there hidden characters here, yes or no, and if so, which ones.
This detector finds invisible and format characters embedded in the text itself. It is precise and complete for that job. It cannot, however, detect statistical watermarks such as Google SynthID-Text, which live in the model's word choices rather than in hidden characters and leave no character trace to scan. If a scan comes back clean, it means there are no hidden characters — not necessarily that the text was written by a human.
Everything runs locally in your browser, so you can analyze sensitive text without uploading it anywhere. Once you know what a text contains, you can leave it as-is or clean it in one click.
Straightforward answers about detecting hidden characters, privacy, and accuracy.
Paste the text into the detector above and click Analyze Text. It scans every character and reports whether any invisible watermark characters are present, how many, of which types, and exactly where they are.
Zero-width spaces and joiners, the byte order mark, bidirectional control codes, variation selectors, tag characters, soft hyphens and non-standard spaces. All of them are invisible or near-invisible to a reader but detectable in the raw text.
No. SynthID-Text is a statistical watermark encoded in word choice, not as characters, so there is nothing for a character scanner to find. A clean result here only means there are no hidden characters in the text.
No. Detection runs entirely in your browser. Your text is never uploaded, stored or logged, so you can safely check private content.
Use the AI Watermark Remover, linked from the results. It strips every invisible character in one click and copies the clean text to your clipboard.
Not necessarily. It means the text has no hidden characters. Detecting authorship is a different problem that character scanning cannot solve.