Unofficial
AI watermark detector
Scan to list hidden characters. Your text stays in this tab.
What this detector finds
It finds characters you cannot see: zero-width spaces, byte-order marks, unusual spaces, and letters that only look Latin. It does not find statistical watermarks such as textGrain, Claude’s watermark, or SynthID-Text, because those live in word choice and need the provider’s secret key.
How to check a passage
- Paste.
- Press Scan.
- Read the list, then press Strip if you want those characters gone.
What a character scan cannot tell you
| This site — free | This site — paid rewrite | Typical zero-width cleaners | |
|---|---|---|---|
| Zero-width characters | Find and delete, in the browser | Same as free | Find and delete |
| Em dash | Optional punctuation cleanup — not called a watermark | Same as free | Often labeled as a watermark |
| textGrain / Claude statistical watermark | Cannot detect | Rewrite changes wording; no result promise | Often imply cleanup is enough |
| Privacy | Text stays in the browser | Rewrite is sent to the model for that request; we do not store it | Often unclear |
When a rewrite is the only thing that changes the wording
A paid rewrite changes the words themselves. It is a separate step from stripping hidden characters. It does not promise how any detector will score the result. See Pricing.
FAQ
What is an AI watermark detector?
A tool that looks for hidden characters in text. This one runs locally and lists each code point it finds.
Can you detect ChatGPT’s textGrain?
No. textGrain is a pattern in word choice. OpenAI’s detector is not public, and this site does not claim to run it.
Does the text leave my browser?
Not for Scan or Strip. A rewrite, when that feature is on, is sent to the model provider to produce new wording. We do not store the text.
Is this affiliated with OpenAI, Anthropic, or Google?
No. Unofficial.
