Anthropic Plans Claude Watermarks for AI Generated Content
Anthropic will add machine readable marks to Claude outputs from new models launched in the EU on or after August 2, 2026. Text will use embedded watermarks, while supported files will use signed provenance metadata.