Anthropic has published where its invisible mark stops working: rewrite every word and it is gone
Anthropic put out a technical account of the watermark it started embedding in Claude's text this month, and it is unusually direct about the limits. Light editing probably will not remove the mark, the company says, but 'a complete rewrite where every word is replaced will'. Detection works poorly on short passages, because fewer word choices mean less signal to read. The mark is sparser in factual writing and in code, where there are fewer ways to say the same thing without making it wrong. And if you hand Claude a document and ask only for grammar and punctuation fixes, the watermark can live only in those few corrections, which may be too few to register. Anthropic also spells out what a positive detection means: that Claude was probably involved in producing the text — not that no human wrote it.
Related stories
Apple Music will tell you when a song was made by a machine — but the uploader decides whether to say so2026-08-21Stripe has just paid $7.5 billion for a model router. Days later Ramp built one and is giving it away until January.
2026-08-21Meta's assistant is now a Mac app that reads your screen and types into any window — and what it sees can train the model
2026-08-21One click on a news site now tells Google to show you more of it — and you stay on the page you were reading
2026-08-21Adobe will now generate the music, the voiceover and the door slam — and it says the licence covers you
2026-08-21