.png){kind=logo}
Anthropic introduced invisible watermarks for text generated by its Claude artificial intelligence models, extending the marking system worldwide instead of limiting it to the European Union.
The company will embed an imperceptible watermark into text generated by Claude models launched on or after August 2, 2026. It will also attach signed C2PA provenance metadata to generated image files. Models released before August 2 are still being retrofitted.
Anthropic’s decision goes beyond the requirements of European regulations. The company signed the European Union’s transparency code for AI-generated content and is implementing the marking system across markets where supported Claude models are available.
The watermark will be woven directly into the generated text, making it imperceptible to users. It will not affect the meaning, quality, or readability of Claude’s responses. Since the mark is part of the text itself, it is designed to travel with the content when it is copied and pasted elsewhere and may remain through some editing.
The system will operate at the model level, meaning the watermark will remain present regardless of the Claude product or surface used to generate the content.
The marking system will cover Claude, Claude Platform API, Claude Code, Claude Cowork and Claude Tag. Anthropic said the watermarks will also apply when Claude models are accessed through cloud partners including Amazon Web Services, Google Cloud and Microsoft Foundry.
For generated images, supported file formats including .svg, .png and .jpg will carry digitally signed provenance metadata using the C2PA open standard.
Also Read: Meta, OpenAI, Anthropic Report AI Models Taking Unauthorized Actions
The August 2 date is important as watermarking applies at launch to Claude models released on or after that date. Models released earlier are still being retrofitted, and Anthropic did not announce a specific date for completing that process.
The move follows the European Union’s AI transparency requirements, but Anthropic chose to implement the system globally rather than restricting it to European users.
The company is also developing tools that will allow users and third parties to detect the watermarks. Anthropic said technical details about the detection system will be released later. The initiative aims to make AI-generated content easier to identify and improve transparency around the use of Claude-generated material.
