How to remove Claude watermarks and hidden characters
To remove a Claude watermark, paste your text above and click Clean my text. Text generated by Claude carries small giveaways: curly "smart" quotes, em dashes, non-breaking spaces, and, when copied through editors, invisible zero-width and formatting characters.
TextUnmark strips these instantly and shows a report of exactly what it removed. Your words stay identical. Only the machine-style characters go.
Hidden characters
Zero-width spaces, joiners, BOMs, and directional marks are removed.
Em dashes
Claude's favourite dash becomes a comma between words, so the sentence still reads right.
AI buzzwords
Tells like "delve," "leverage," and "seamless" are swapped for plain language.
Private
Nothing is stored or logged. It runs in memory only.
How to remove a Claude watermark, step by step
- Paste your Claude text into the box at the top of this page.
- Click "Clean my text" to strip hidden characters and normalize Claude's typography.
- Copy your clean copy. Review the change report, then copy the result with one click.
What is a Claude watermark?
A Claude watermark is a signal that marks text as generated by Anthropic's Claude AI. Two different things get called a "Claude watermark," and it helps to know which one you are dealing with.
1. Character-level marks and typography. These are invisible Unicode characters, smart quotes, and em dashes. This is exactly what TextUnmark cleans, instantly and reliably.
2. A statistical watermark. Anthropic embeds this newer signal in word choice itself. Because it lives in the wording rather than in hidden characters, no find-and-replace cleaner can remove it. That requires genuinely rewriting the text. We tell you this plainly, because a tool you can trust is worth more than an empty promise.
Frequently asked questions
What is a Claude watermark?
A Claude watermark is a signal that marks text as generated by Anthropic's Claude AI. There are two kinds: character-level tells such as smart quotes, em dashes, and invisible Unicode characters, and a newer statistical watermark baked into Claude's word choice. TextUnmark removes the character-level tells. The word-choice watermark can only be removed by genuinely rewriting the text.
Does Claude add watermarks to its text?
Yes. Claude output commonly includes typographic tells like curly quotes, em dashes, and non-breaking spaces, and text copied through editors can pick up zero-width and formatting characters. Anthropic has also begun embedding a machine-readable statistical watermark in Claude text to meet AI transparency rules such as the EU AI Act.
How does this watermark remover work?
You paste your text and click Clean text. TextUnmark scans every character, removes invisible Unicode characters, and converts smart quotes, em and en dashes, non-breaking spaces, and ellipses to plain equivalents. It all happens in memory, with a report of exactly what changed. Your words are never altered.
Is it legal to remove AI watermarks?
Cleaning invisible characters and typography from text you own or are allowed to use is generally fine, and that is what TextUnmark is built for. That said, laws and platform rules vary by country and context, and removing a watermark to pass AI content off as human where disclosure is required could break rules or terms of service. This is not legal advice, so check the rules that apply to you.
Do other AI tools like ChatGPT use watermarks?
AI providers are increasingly adding watermarks and provenance signals to generated text and images, driven by regulations like the EU AI Act. Support and methods vary by tool and keep changing. TextUnmark works on text from any source, and we add dedicated pages for specific tools as their watermarking rolls out.
How do I remove em dashes from Claude text?
Paste your Claude text above and click Clean my text. An em dash between words becomes a comma so the sentence still reads correctly, a range between numbers becomes a hyphen, and any other dash becomes a plain hyphen. The report tells you how many were changed.