Anthropic embeds invisible watermarks in all Claude text to meet EU AI Act Article 50
On August 11 Anthropic announced all Claude models released on or after August 2 carry invisible text watermarks that persist through copy-paste, applied globally.
On August 11, Anthropic announced that all Claude models released on or after August 2 carry default AI-generated content marking: every generated text carries an invisible watermark and every supported file carries digitally signed provenance metadata. The move follows Anthropic's signing of the EU AI Act Article 50(2) Code of Practice on transparency of AI-generated content.
Article 50 of the EU AI Act took effect on August 2, requiring providers of generative AI systems to mark synthetic outputs in a machine-readable format. The accompanying Code of Practice on Transparency of AI-Generated Content was implemented at the same time. By the end of July, around 190 organizations had signed the code, including Anthropic, Google, Meta, Microsoft, and OpenAI.
Claude uses two marking methods. For text, an imperceptible watermark is woven directly into the generated text without changing meaning, quality, or readability; the watermark travels with the content when copied and pasted and may persist through some editing. For files, supported formats like .svg, .png, and .jpg carry digitally signed provenance metadata following the C2PA open standard.
Coverage is global across the entire product line: API, Claude, Claude Code, Claude Cowork, and Claude Tag, including cloud customers accessing Claude via AWS, Google Cloud, and Microsoft Foundry. Anthropic said it is developing companion detection tools for users and third parties to verify whether text or files carry Claude marks. The company also explicitly noted limitations: heavy rewriting, translation, or mixing with other text may erase the watermark; detection only indicates the content was processed by Claude, not that Claude authored the source material.