Claude to start watermarking AI-generated content

Anthropic has announced that its Claude AI model will begin watermarking AI-generated content to comply with EU regulations. In a support document, the company explained that it has adhered to Article 50(2) of the EU AI Act's Code of Practice on Transparency of AI-Generated Content. Claude models launched in the EU from August 2, 2026, will have watermarking features enabled at launch, including text and generated files.

However, watermarks won't be visible unless users actively search for them. Text files will have the watermark embedded in the content itself, while files like images will follow the C2PA open standard. Anthropic emphasized that the watermarking system isn't foolproof and may not detect heavily edited, paraphrased, or translated text.

The company also noted that existing models launched before August 2, 2026, will eventually be updated to support watermarking.