Anthropic Introduces Invisible Watermarks for Claude to Comply with EU AI Act

xiaohu · x · 2026-08-11

Anthropic has announced a content marking scheme for Claude, embedding invisible watermarks in generated text and adding signed provenance info to image files. This feature is applied at the model level across all Claude products.

This move primarily complies with the transparency requirements of Article 50 of the EU AI Act, effective August 2. Around 82 providers, including Google and OpenAI, have signed the related code of practice. Anthropic emphasized that while the watermark is machine-detectable, it cannot definitively prove whether content is AI-generated.

Related event: Claude Rolls Out Invisible Watermarks and Metadata, Sparking Compliance and Tech Debates(115 posts)→

Original post →

More from Models

Models channel →