According to CoinMeta and Wall Street Lunch, Anthropic has announced that in the new models to be launched in the European Union, watermarks will be embedded in all content generated and edited by AI. This change stems from Anthropic signing into Article 50(2) of the EU AI legislation regarding transparency of AI generated content, which will apply to all new models launched in the EU from August 2nd onwards. Anthropic states that the generated text will carry embedded watermarks, and the generated files will include source metadata with digital signatures. The watermarks will be applied to all supported Claude models and will be present in Claude products worldwide.