What this tool cannot do
We remove characters. We do not remove statistical watermarks.
Modern AI watermarks — Google's SynthID-Text and the watermark Anthropic ships for Claude — are statistical. They live in which words the model chose, not in any character added to the text. Anthropic's own documentation states plainly that nothing is added and there are no hidden characters. Stripping invisible characters therefore has no effect on them at all, and any tool that claims otherwise is selling you something that cannot work. What we do is real and verifiable: we find and remove invisible characters, decode hidden payloads so you can see them, and normalize typography — and we show you a full report of every single change.
What we can do
- ✓Remove zero-width and invisible Unicode characters (U+200B/C/D, U+2060, U+FEFF, U+00AD and more).
- ✓Find and decode hidden-text payloads in the Unicode Tags block — the channel used for prompt injection.
- ✓Identify StegCloak-style and base-N zero-width steganography, and show you the decoded message.
- ✓Flag bidirectional controls that make text render in a misleading order.
- ✓Replace unusual spaces with ordinary ones instead of deleting them, so words never merge.
- ✓Normalize smart quotes, dashes, ellipses and primes to plain ASCII, when you ask for it.
- ✓Flag look-alike letters from other scripts (Cyrillic а, Greek ο) without silently replacing them.
- ✓Show a complete report of every character changed, with its position, name and code point.
What we cannot do
- ✗Remove Google's SynthID-Text watermark. It is statistical — it lives in which words the model chose, not in any character. Cleaning characters has no effect on it.
- ✗Remove Anthropic's watermark for Claude. Anthropic's own documentation states that nothing is added to the text and there are no hidden characters.
- ✗Remove “the ChatGPT watermark.” There isn't one. OpenAI built a text watermarking system but never deployed it. The narrow no-break space (U+202F) that went viral is a formatting artifact, not a mark — it encodes nothing.
- ✗Guarantee that text will pass an AI-detection service. Those tools classify writing style and sentence structure; they do not look at invisible characters.
- ✗Make text undetectable as AI-generated. No character-cleaning tool can do this, and we will not pretend otherwise.
This page was written from the primary sources, not from a competitor's marketing. If you believe any claim here is wrong, tell us and we will correct it.