Roughly three-quarters of newly published web pages now carry some machine-written text, according to a 2025 Ahrefs analysis of around 900,000 pages. This figure doesn’t account for offline documents. It also only covers text. Generative AI now produces music, illustrations, software, photographs and increasingly convincing video. The problem is that much of it is becoming difficult to distinguish from human-created work, raising transparency and security concerns.
The European Union's answer to these concerns is regulation. Under Article 50(2) of the EU AI Act, providers of generative AI systems are required to make all generated content detectable as artificially generated or manipulated.
Anthropic, the company behind Claude, has signed the bloc's Code of Practice on Transparency of AI-Generated Content, one of around 190 signatories, and now plans to watermark Claude's output across its ecosystem. According to the company, text will carry an invisible statistical watermark, while supported visual formats (such as PNG, JPG, or SVG) will receive signed metadata noting that Claude was involved.
Generative AI has spread into writing, music, art, software, and almost anything digital. For many, it has made work faster and cheaper. Conversely, it has also raised serious concerns about irresponsible use and security. Using AI to fully generate academic essays or in content contracts where AI is explicitly disallowed are valid concerns. Then there are less serious cases where people discover that the impressive song or video they admired was deceptively AI-generated all along.
However, the much bigger issue is around security. AI-generated content has become so powerful that it's becoming increasingly difficult to identify. Of course, the video of a muscular cat moonwalking across the ocean doesn’t count. But a video of a public figure calling for violence or a fake official document can have grave consequences.
These are some of the issues that watermarking and identification can help resolve. Anthropic says its marking applies to output from supported models everywhere Claude runs, including the Claude Platform API, the Claude apps, Claude Code, Claude Cowork and Claude Tag. The scope reaches third-party products built on Claude's API, as the watermark rides inside the output rather than any single interface.
Models launched on or after August 2, 2026 support marking at launch, while the company is working to add it to older models over the coming months under a transition period the EU law allows. Although the requirement is European, Anthropic is applying watermarking worldwide from the start, saying it does not yet have a durable way to limit it by region.
So, how does the marking work? For text, Anthropic uses a version of Google DeepMind's SynthID-Text method. It does not rely on hidden characters or metadata attached to a document. Instead, the watermark is created through the words Claude chooses while generating text. When generating text, large language models pick each next word from a set of plausible candidates. In many situations, several words could continue a sentence equally well, so the choice is normally settled at random.
With SynthID-style watermarking, a secret key and the preceding text influence that choice. Across enough words, those tiny choices build up into a statistical pattern. So while the text still reads naturally, it now carries a signature that a detector with the appropriate key can examine to determine whether its word-selection pattern resembles Claude's watermarked generation. Pretty nifty. Anthropic says the technique adds no characters, needs no extra tokens, doesn't change output quality, and carries nothing that identifies a user, an organization, or a chat.
For supported files, Anthropic uses a more conventional technique. When Claude generates formats such as a PNG, JPG, or SVG, it attaches a cryptographically signed note in the file's metadata under the open C2PA standard. Nothing in the image itself changes. The credential records that Claude made or processed the file and holds no personal information. A compatible reader can inspect that credential and determine that Claude processed the file. Because the information is cryptographically signed, it can also help reveal whether the credential or file has subsequently been tampered with.
While Anthropic has figured out how to mark its text output, it's still figuring out how to detect those markings. Anthropic says a watermark-detection API is coming but has not finalized how it will work. For files, any C2PA-aware tool can read the file credentials, with the company promising its own checker.
Now, Anthropic acknowledges that its marks are signals, not definitive proof of AI-generation. A detected watermark only shows that Claude was likely involved at some point, not that a human did not write it. Because the watermark attaches to any words Claude chooses, it cannot separate authorship from assistance. Anthropic notes that it "cannot distinguish 'Claude wrote this' from 'Claude heavily edited this,'" and that a translation Claude produces is fully watermarked because every word is its choice.
On the detection side, the company also concedes that while light editing probably will not remove the watermark completely, "a complete rewrite where every word is replaced will," at which point, it argues, the text is barely AI-generated anyway. Text that has been heavily rewritten, paraphrased, translated, or mixed with other writing may lose enough of Claude's statistical pattern to become undetectable. In practical terms, someone deliberately trying to conceal Claude's involvement could simply rewrite the output substantially.
Source: Claude
