Anthropic to Add Watermarks to All Claude Outputs Under EU AI Act

jedisct1 · x · 2026-08-11

Complying with the EU AI Act's transparency requirements, Anthropic plans to embed watermarks and provenance metadata into Claude's outputs.

Timeline: New models launched in the EU on or after August 2, 2026, will support machine-readable marks from day one, with existing models updated during a transition period.
Scope: The marking will apply globally across all Claude platforms, including the API, Claude.ai, and Claude Code.
Detection: Anthropic will provide documentation and tools to help users and third parties detect these watermarks.

More from Models