Does Anthropic really watermark Claude's text?

Yes, since 11 August 2026. Every Claude model launched on or after 2 August 2026 embeds an imperceptible mark in generated text, applied at the model level so it is present on every Claude surface, and applied worldwide rather than only in the EU. The driver is EU AI Act Article 50, which became enforceable on 2 August 2026. Anthropic states the mark travels through copy-paste and may persist through some editing.

Can the watermark actually be removed?

Anthropic's own documentation says the mark may not survive heavy editing, paraphrasing, translation, mixing with other content, or very short passages. Paraphrasing is on that list, and the rewrite here is a paraphrase pass. But Anthropic has not published its detection mechanism, so nobody outside Anthropic - including us - can currently verify whether a mark is present or gone. We will not claim a result we cannot demonstrate. When a detector is published we will test against it and publish what we find.

What does a detected mark actually prove?

Less than people assume. Anthropic states a mark indicates content may have been processed by Claude - not that Claude authored it. Text you wrote yourself and asked Claude to proofread carries the same mark as text Claude drafted from scratch. The mark cannot distinguish those two cases.

What are the two layers you clean?

Character-level and statistical. The character layer is zero-width codepoints, Unicode tag characters, variation selectors, exotic spacing and typographic punctuation - removed by exact codepoint matching, verifiable by re-scanning, no model involved. The statistical layer is the vocabulary and cadence patterns, plus whatever the watermark contributes; that can only be addressed by rewriting.

Can invisible characters really be hidden in text?

Yes, and they survive ordinary copy-paste. Zero-width characters, Unicode tag characters and variation selectors all render as nothing while remaining fully present in the data. Run the free scan on any text and you will see the exact count.

Does the rewrite change my meaning?

It is instructed not to. Facts, figures, names, dates, quotations, citations, URLs, paragraph structure and claim strength are all held fixed, and any sentence without a marker is returned verbatim. Read the result before you use it — that is true of any tool that touches your writing.

Will this defeat a specific AI detector?

We make no claim about any named detector, and no claim to verified watermark removal. Detectors are proprietary, they change without notice, and they disagree with each other. Anthropic's watermark has no public detector at all yet. What we state precisely: the character-level markers are removed exactly and verifiably, and the wording markers in your scan report are what the rewrite targets.

Why is the scan free?

Because it costs us almost nothing. It is exact character matching, not a model call, so there is no reason to charge for it. You only spend words when you ask for an actual rewrite.

What happens to my text?

It is processed to produce your result and saved to your history so you can retrieve it later. You can delete any entry. It is never used to train a model.

How is usage counted?

By word, against the input text. Scans are free and never counted. Each plan also caps the words in a single request, so long documents need a higher tier or splitting up.

Can I use this on text I wrote myself?

Yes, and plenty of people do. Human writing picks up invisible characters from PDFs, CMS editors and word processors, and the typographic punctuation cleanup is useful on its own. The scan will tell you whether there is anything there.

Claude vs ChatGPT vs Gemini: which AI watermarks text

Where each provider stands on text watermarking as of August 2026, what is verifiable, and what it means for output you already have.

Which providers watermark · 2 min read

Text watermarking went from one provider to two in August 2026. Here is where each stands, and what is actually checkable.

The current state

Provider	Text watermark	Public detector	Notes
Anthropic (Claude)	Yes, since Aug 2026	No — "forthcoming"	Models launched on/after 2 Aug 2026; older models in progress. Applied at model level, worldwide
Google (Gemini)	Yes — SynthID-Text	Key-holder only	Published in Nature, Oct 2024. Best-documented implementation
OpenAI (ChatGPT)	No	n/a	Researched and reportedly built, not deployed

Anthropic: newest, least documented

Claude embeds an imperceptible mark in generated text. Applied at the model, so every surface inherits it — app, API, or anything built on top. Applied globally rather than only in the EU, though EU AI Act Article 50 (enforceable 2 August 2026) is the driver.

Anthropic states the mark travels through copy-paste and may persist through some editing, and that it may not survive heavy editing, paraphrasing, translation, mixing with other content, or very short passages.

Detection documentation has not been published, so nobody outside Anthropic can currently check for it.

Google: oldest, best understood

SynthID-Text biases token sampling using a secret key so that a paired detector can identify Gemini text statistically. Because Google published the method, its limits are documented rather than guessed: it degrades under paraphrasing and translation, and needs a reasonably long passage.

Only Google can detect it. It is not a public checksum.

OpenAI: researched, shelved

OpenAI built a text watermarking method and, according to reporting in 2024, held it back over concerns about the effect on users — non-native English speakers in particular — and how easily paraphrasing defeats it.

It also shipped and then retired an AI-text classifier in July 2023 for low accuracy. That was a detector, not a watermark.

The pattern across all three

Every implementation shares one weakness, because it is inherent to the approach rather than to any vendor: the signal lives in which words were chosen. Change the words and it degrades.

That is why:

Google documented paraphrasing as a limitation.
OpenAI cited it as a reason not to ship.
Anthropic lists paraphrasing among the things its mark may not survive.

It also means watermarking reliably catches unedited copy-paste and reliably misses anyone who edits — which is backwards from a compliance perspective, since it penalises the careless rather than catching the deliberate.

What this means for text you already have

Claude output from a current model is marked. Unverifiable by you.

Gemini output may carry SynthID. Unverifiable by you.

ChatGPT output carries no deployed watermark.

All three carry stylistic fingerprints — em dashes, curly quotes, the U+2026 ellipsis, signature vocabulary — and those are exactly countable. Scan any text and you will get the numbers, whichever model produced it.

Sources: Anthropic Help Center · TechCrunch, 11 Aug 2026

Claude vs ChatGPT vs Gemini: which AI watermarks text

The current state

Anthropic: newest, least documented

Google: oldest, best understood

OpenAI: researched, shelved

The pattern across all three

What this means for text you already have

Related

Does ChatGPT watermark its text?

What SynthID-Text is and how statistical text watermarking works

Does Claude's API output carry the watermark?