Anthropic has started adding invisible watermarks to text generated by its newer Claude models, extending the measure worldwide as it prepares to meet requirements under the European Union’s
Anthropic has started adding invisible watermarks to text generated by its newer Claude models, extending the measure worldwide as it prepares to meet requirements under the European Union’s AI Act.
Claude models launched from 2 August will carry an imperceptible signal at the model level. The watermark does not change the text’s meaning or readability, and it travels with the text when copied and pasted. Anthropic said the mark “may persist through some editing.” The company has signed the EU’s Code of Practice on AI-generated content, and it said the markings will apply worldwide, not only to users in the EU.
The signal will apply across the Claude app, the API, Claude Code, Claude Cowork and Claude Tag. Customers accessing supported models through Amazon Web Services (AWS), Google Cloud and Microsoft Foundry will also receive marked output.
Files are handled differently. Supported file formats, including common image types, will carry signed provenance metadata based on the Coalition for Content Provenance and Authenticity (C2PA) standard, rather than an embedded text watermark.
Read also: What OpenAI’s unlimited ChatGPT rollout actually fixes for African users
Older Claude models will not immediately receive the marking system. Anthropic said it is working to extend the capability to those models during the EU AI Act’s transition period.
The company said heavy editing, paraphrasing, translation, or mixing Claude’s output with other writing can make the watermark undetectable. It added that detecting a watermark does not confirm the chatbot originally authored a piece of text, since activities such as proofreading or translating existing work with Claude can also leave a mark.
Anthropic said it plans to publish technical details to help users and third parties detect its watermarks, with further documentation expected as the marking and detection systems develop.
Anthropic is the second major AI lab to introduce text watermarking. Google DeepMind announced in 2024 that it was watermarking text and video generated through the Gemini app and web experience using its SynthID technology, having released an image-watermarking version of the tool in 2023.
