Explained: Claude's Watermark Policy to Comply with EU AI Act Globally

talkaboutdesign · x · 2026-08-11

Anthropic detailed its implementation plan for marking AI-generated content in Claude, fulfilling the transparency requirements of Article 50(2) of the EU AI Act.

Starting August 2, 2026, new Claude models launched in the EU will support machine-readable marks from day one, including embedded text watermarks and digitally signed metadata. These marks will apply globally across Claude's API, web interface, and integrated tools. Anthropic also committed to helping third parties detect these marks and is working to add support for existing models.

Related event: Anthropic to Embed Invisible Watermarks and C2PA Signatures in Claude Outputs from Aug 2026(14 posts)→

Original post →

More from Models

Models channel →