Anthropic watermarks Claude everywhere because it cannot yet watermark only Europe
Anthropic says the mark exists to satisfy the EU AI Act and that it has no durable way to scope it by region, so every customer gets it. OpenAI's support page says text watermarks are coming too.
The Board Room · Leadership desk
What happened
- Anthropic told Business Insider that the watermark has applied to Claude-generated text globally on supported models since August 2.
- Anthropic told Business Insider it is adding the watermark in order to comply with the European Union's AI Act.
- Anthropic published a blog post on Friday that defends and explains the watermarking technology and outlines its limitations, after concerns from Claude users.
- Anthropic said in its post that it is "applying watermarking globally at launch because we don't yet have a durable way to scope it by region."
- Anthropic's post says its AI provider peers will have to watermark text as well, which Business Insider described as a likely nod toward OpenAI.
Why it matters
Anthropic has been embedding an imperceptible watermark in Claude-generated text on supported models globally since August 2, and it told Business Insider the reason is compliance with the European Union's AI Act [1][2]. In a blog post on Friday defending the feature, the company said it is "applying watermarking globally at launch because we don't yet have a durable way to scope it by region" [21][7].
That is the clearest case yet of an EU obligation shipping as a worldwide product default. The trigger is not where the customer sits or which law binds them; it is which model handled the text [25].
Anthropic's post also said its provider peers will have to do the same, which Business Insider read as a likely nod toward OpenAI [9]. OpenAI did not immediately respond to Business Insider's request for comment, though a support page updated two weeks earlier says it also plans to add watermarks to text [11][13]. Google already watermarks AI-generated content using SynthID, and OpenAI uses SynthID for supported images and audio [23][24].
The technique, drawn from a 2024 Google DeepMind paper, works on word choice [3]. Where a model could pick "joyful" or "cheerful" at random, the watermark biases that selection so that holders of a key can detect the pattern [4]. There are no hidden characters, unusual fonts, or secret text, and Anthropic says readers will not be able to tell the difference [5]. The key has not been released; Anthropic says it will offer a free API so users and third parties can check text themselves [6][27].
The limits matter more than the mechanism for anyone relying on this. Anthropic's post says that with its key, "one can only answer the question 'What is the likelihood this was partly written by Claude?'" [8]. Longer passages are easier to detect because more word choices were made, factual passages get fewer marks to protect accuracy, and code gets fewer still because a wrong term can stop software running [10][12]. Light editing probably will not remove the mark; a complete rewrite will, at which point Anthropic notes it is arguable whether the text is AI-generated at all [14].
The objections are governance objections, not aesthetic ones. Investor Bill Gurley said that if only Anthropic can identify the mark, the company becomes "judge, jury, and prosecutor" [18]. Former Microsoft executive Steven Sinofsky raised data retention and "your right to private thoughts free of a digital trail" [19]. Trainer John Crickett questioned whether watermarked code complicates copyright claims where authors cannot show sufficient human input [20]. Klick's Simon Smith warned that lightly assisted work, such as a grammar check, could be read as AI-authored [22]. Anthropic's answer is that a watermark shows Claude processed the content, can survive proofreading, translation or summarising, and does not change a user's rights or ownership under its terms [15][16].
Commercially, some customers told Business Insider they cancelled Claude subscriptions over the watermark, and Anthropic told the publication it has not seen an uptick trend in cancellations [28][29].
Watch three things. Whether "durable" regional scoping arrives, because that determines if global marking was a transitional constraint or the new baseline. Whether the free detection API and key actually ship, since the credibility of the whole scheme rests on third parties being able to check [6]. And whether provenance labels hold up operationally: X's "Made with AI" tag appeared on outgoing White House Press Secretary Karoline Leavitt's resignation post on Wednesday, then disappeared [26]. Anthropic says older models get the watermark in the coming months, so procurement teams have a short window to fix ownership and disclosure language in contracts before the mark is everywhere [17].
Clarity's read
What the record supports and how the coverage leans. The claims behind it follow.
Reality
- Evidence60
- Adoption45
- Hype gap+15
- Incentives60
- Confidence62
Perspective Coverage
3 publishers- Builder
- Builder 40%
- Operator
- Operator 42%
- Investor
- Investor 18%
Claim ledger
Ranked by verification strength, evidence, and original report placement.
- [1]
Anthropic told Business Insider that the watermark has applied to Claude-generated text globally on supported models since August 2.
- [2]
Anthropic told Business Insider it is adding the watermark in order to comply with the European Union's AI Act.
- [3]
Anthropic cites a 2024 Google DeepMind paper that established the watermarking technique it will apply to Claude outputs.
- [4]
The watermark works through word choice: where a model could randomly select between synonyms such as "joyful" or "cheerful", the watermark acts as a different type of random number generator, and a holder of a key can detect whether text follows Claude's subtle patterns of choice.
- [5]
The watermark does not include hidden characters, unusual fonts or secret text, and Anthropic says the difference will not be distinguishable to readers.
- [6]
Anthropic has not released the key, and says it will offer a free API allowing users and third parties to check text for Claude's watermark themselves.
- [7]
Anthropic said in its post that it is "applying watermarking globally at launch because we don't yet have a durable way to scope it by region."
- [8]
Anthropic's post says the key points to Claude's involvement rather than AI in general: "Using our key, one can only answer the question 'What is the likelihood this was partly written by Claude?'"
- [9]
Anthropic's post says its AI provider peers will have to watermark text as well, which Business Insider described as a likely nod toward OpenAI.
- [10]
Longer passages will be easier for the key to recognise as Claude-generated because the model made more word choice decisions, and "factual passages" where wrong choices might jeopardise accuracy will receive fewer marks.
- [11]
OpenAI did not immediately respond to Business Insider's request for comment.
- [12]
Anthropic writes that a watermark would not be applied as often in code because code is exact and the wrong term might mean the software cannot run correctly.
- [13]
An OpenAI support page updated two weeks before the report says the company also plans to add watermarks to text.
- [14]
Anthropic's post says: "Light editing probably won't remove the watermark completely; a complete rewrite where every word is replaced will," adding that in the latter case it is arguable whether the text can still be described as AI-generated.
- [15]
Anthropic said a watermark shows that Claude processed text, not necessarily that it wrote it, and that the mark can remain after Claude has proofread, translated or summarised content.
- [16]
Anthropic says the watermark appearing indicates that Claude processed the content or file and does not change a user's rights or ownership of the text under its terms.
- [17]
Anthropic's post says watermarks will first be applied to new Anthropic models and then to older ones in the coming months.
- [18]
Tech investor Bill Gurley raised the concern that only Anthropic would be able to identify the watermark, which he said would make the company "judge, jury, and prosecutor."
- [19]
Former Microsoft executive Steven Sinofsky wrote in an X post on Tuesday: "The real issue is data retention and your right to private thoughts free of a digital trail."
- [20]
Software engineer trainer and coach John Crickett questioned in an X post on Tuesday whether a watermark on AI-generated code could complicate copyright claims because authors would not be able to show sufficient human input.
- [21]
Anthropic published a blog post on Friday that defends and explains the watermarking technology and outlines its limitations, after concerns from Claude users.
- [22]
Simon Smith, who leads generative AI at digital health agency Klick, said in a Tuesday X post that one of his concerns was that lightly edited AI work, such as a grammar check run through Claude, could be flagged as AI-authored.
- [23]
Google uses its SynthID technology to watermark AI-generated content.
- [24]
OpenAI uses SynthID for supported images and audio.
- [25]
Because Anthropic applies the mark globally on supported models for EU AI Act compliance and says it lacks a durable way to scope by region, the condition that determines whether a customer's text is watermarked is the model used, not the customer's jurisdiction.
- [26]
X adds a "Made with AI" tag to content it determines to be AI-generated or manipulated; the tag appeared on the resignation post of outgoing White House Press Secretary Karoline Leavitt on Wednesday, but later disappeared.
- [27]
An Anthropic spokesperson said the watermark "doesn't change the meaning, quality, or readability of Claude's responses."
- [28]
Some customers told Business Insider they cancelled Claude subscriptions because of the new watermark.
- [29]
Anthropic told Business Insider it has not seen a trend of an uptick in cancellations since announcing the watermark.
Sources
3 independent publishers whose own reporting we read for this story.
- businessinsider.comTechies have concerns about Claude's hidden watermark. Anthropic has some answers.
5 articles · August 18, 2026
- forbes.comClaude Will Now Leave A Watermark On Everything It Writes. What Does That Mean?
5 articles · August 20, 2026
- theguardian.comClaude to start watermarking AI-generated text – but will it make quality worse?
1 article · August 17, 2026