What happened
Anthropic published a detailed explanation of how Claude’s text watermarking works on August 14. The company signed the EU’s Code of Practice on Transparency of AI-Generated Content in July 2026, and Anthropic says the EU AI Act requires providers serving the EU market to mark AI-generated content.
The watermark works by changing how Claude picks between equally good word choices, not by hiding extra characters or attaching a tag to the text you’d notice. For low-stakes word choices, it uses a hidden key and preceding words to produce a detectable statistical pattern. Read on its own, a watermarked response looks completely normal. Anyone holding the matching key can run the text through a checker and estimate the likelihood that Claude contributed to it.
Anthropic based this on SynthID-Text, a watermarking method Google DeepMind published in 2024 in Nature, a scientific research journal. It’s rolling out globally now for new Claude output, with older models getting the same treatment over “the coming months,” per Anthropic’s post.
Why it barely touches your code
Anthropic says watermarking acts on choices where alternatives preserve the intended output. Exact code tokens leave less room for that pattern. Code has to follow strict syntax rules, and a variable name has to match wherever it is used. Anthropic says those constraints leave code with less watermarking than many other forms of text. Comments, which are freeform explanatory text and not code the computer executes, may carry more of it, but the company says the effect on the actual code produced is negligible.
Keep checking generated code normally: watermarking does not establish that a program is correct.
Anthropic also addressed whether editing removes the watermark. Light editing “probably won’t remove the watermark completely,” but a full rewrite where every word changes will. That’s mostly relevant to prose, not code, but it’s worth knowing if you use watermark results as evidence of Claude’s involvement.
Who should care
This matters most if you ship AI-generated prose alongside your product: release notes, marketing copy, documentation, support replies, or anything a reader might later want to verify as human or AI-written. It matters less for pure code output, since Anthropic’s own explanation says the effect there is negligible.
It’s also worth knowing if you’ve seen chatter about this online. Some Claude users have publicly objected to watermarking, and a few have said they’re canceling subscriptions over it, framing it as Anthropic tracking them. Anthropic’s response is that the watermark doesn’t change output quality or content, only a hidden statistical pattern in word choice, and that it exists to meet a specific EU legal requirement rather than to monitor individual users.
What builders should do next
If a contract requires human authorship, check the actual requirement and keep evidence of your writing process. Removing a watermark does not prove text was written entirely by a person.
Anthropic’s September 1 update says its detection API is in private preview for eligible organizations, with broader access planned. An API lets software request a check automatically.
Detection needs the matching key and enough text for a useful statistical result. It estimates whether Claude contributed, not whether another AI was involved or whether the text is human-written.
For code, the practical takeaway is simpler: nothing changes about how you review, test, or ship what Claude writes for you. For prose, interpret detection as evidence of likely Claude involvement rather than a verdict on authorship.
End of article