How to remove Claude's watermark
Five methods compared with a plain verdict on each. Stripping invisible characters: no effect, because Claude inserts none. Light editing: minimal effect, per Anthropic's own answer. Local synonym rewriting: partial effect, measurable only as the percentage of wording changed. Open-source cleaners such as claude-watermark-cleaner: genuine effect, running rewrite passes until roughly seventy per cent of token sequences are disrupted. Full rewriting through a non-Claude model: most effective, and also the most work.
The honest summary is not that nothing works. Rewriting genuinely works, because the mark is carried by word sequences. What nobody can do is verify the result, because the detector is private.
Frequently asked questions
What actually removes Claude's watermark?
A substantial rewrite. Because the mark is carried by the sequence of word choices, changing most of those choices disrupts it. Character-level cleanup does nothing at all.
Do watermark remover tools work?
Tools that only strip invisible characters do nothing to a statistical watermark. Tools that genuinely rewrite the prose do disrupt it, though none can confirm the result.
Is removing an AI watermark legal?
This is not legal advice. Removing a mark does not remove a disclosure obligation you may have to a school, employer, client or regulator, and under the EU AI Act transparency rules that obligation exists independently of the watermark.
How to get rid of a Claude watermark?
Rewrite the text substantially. Claude's watermark is carried by which words the model chose, so replacing most of those choices is the only thing that disrupts it. Deleting invisible characters achieves nothing, because Anthropic states plainly that no characters are added.
Does Claude AI have invisible watermarks?
Yes, but not invisible characters. Since 2 August 2026 Claude applies a statistical watermark that biases word selection, which is invisible in the sense that you cannot see it — there is nothing hidden in the text to find or delete.
Can I ask AI to remove a watermark?
Asking Claude to remove its own watermark does not work, because the mark is applied during generation rather than added afterwards. Asking a different model to rewrite the text does work, since the rewrite replaces the watermarked word sequence with an unwatermarked one.
How to trick AI into removing watermarks?
There is no prompt trick. The watermark is applied by the sampling process itself, not by an instruction the model can be talked out of. The only reliable method is rewriting the text, either yourself or through a different model.