Anthropic Incorporates Invisible Watermarks into Claude AI Text in Compliance with EU Regulations
Anthropic, a leading AI company, has recently announced its decision to embed invisible, machine-readable watermarks into all text generated by its flagship Claude AI chatbot. This move is a direct response to the EU AI Act’s Transparency Code, which came into effect on August 2, 2026. The code mandates AI companies to mark AI-generated or edited content in a manner detectable by other systems, thereby enhancing transparency.
The watermarks, while being imperceptible to human readers, will not compromise the quality or readability of the text. Importantly, the embedded signal is designed to travel with the text when copied and pasted and is resilient to light editing. Anthropic has ensured that all Claude models released post-August 2 will automatically incorporate this technology. Additionally, the company has committed to retrofitting older models during the regulatory transition period.
For file marking, Anthropic utilizes the open C2PA (Coalition for Content Provenance and Authenticity) standard to attach digitally signed provenance metadata. The company also plans to release technical documentation, enabling third parties to independently verify whether content was processed by Claude. This announcement has wider implications for the AI industry, as Anthropic confirmed that watermarking will be implemented worldwide, not just within the EU, wherever Claude is offered. Other AI platforms, including music generator Suno and newsletter platform Substack, have echoed similar commitments in recent weeks.
Source: TechCrunch – Anthropic says it will watermark text generated by its AI models