Anthropic introduces innovations in marking artificially created content
Anthropic announced the use of new methods for marking content created and processed by its Claude model using watermarks and metadata. This approach aims to enhance the transparency of AI content identification.
Anthropic officially announced the implementation of invisible watermarks in texts and metadata in digital files generated by Claude. New Claude models to be released in the EU from August 2, 2026, will automatically be equipped with these markings. The company is also working on adding such markings to previous models.
Text materials will receive an invisible watermark that will be embedded within the text and will not affect its content. It will be safely preserved during any text operations. For graphic files such as .svg, .png, .jpg, Claude will add metadata based on the C2PA standard to verify the origin of the content.
The marking will be available across various platforms, including Claude Platform, Claude Code, Claude Cowork, and Claude Tag. Additionally, the technology will work in conjunction with AWS, Google Cloud, and Microsoft Foundry. Anthropic is preparing to release watermark detection tools to help users and developers identify AI content. However, the company warns that marking does not prove that all the content is AI-generated, as it may disappear during significant changes or file conversions.
Other tech giants are also actively integrating similar systems. Google indicates whether advertising content is created using AI, and Meta already uses the AI info tag in advertising. Additionally, in 2023, leading AI companies like OpenAI, Alphabet, and Meta committed to the White House to implement watermarks on AI content.
In 2026, LinkedIn and the American Academy of Motion Picture Arts and Sciences also introduced their rules regarding AI content. LinkedIn implemented measures against fake content, and the Academy banned AI-created scripts and actors from participating in the Oscars. Similar initiatives are supported by other platforms, such as the music streaming service Deezer.
| Company | Initiative | Date |
|---|---|---|
| Anthropic | Watermarks and metadata | From August 2026 |
| AI advertising tag | 2023 | |
| Meta | AI Info in advertising | 2023 |
| Fighting fake content | May 2026 | |
| American Academy of Motion Picture Arts and Sciences | New Oscar rules | 2026 |




