Anthropic Details Claude's Invisible AI Text Watermarking

AI | The Verge· August 17, 2026 View original

Key takeaways

  • Anthropic is implementing invisible watermarks for Claude-generated text.
  • The technology is based on Google DeepMind's SynthID-Text approach.
  • This move ensures compliance with the EU AI Act's transparency rules.
  • AI-generated content will require machine-readable marks for authenticity.

Who benefits

AI DevelopmentMedia & PublishingLegal & ComplianceMarketingCybersecurity

Summary

Anthropic has clarified its plan to apply invisible watermarks to text generated by Claude, using a version of Google DeepMind's SynthID-Text approach. This initiative, along with C2PA support for images, aims to comply with the EU's AI Act transparency requirements for synthetic content.

Anthropic recently provided details on how it intends to implement invisible watermarking for text produced by its Claude AI models. The company will utilize a method similar to Google DeepMind's open-source SynthID-Text technology, which embeds detectable patterns within text by subtly adjusting wording probabilities. This watermarking capability, alongside support for C2PA standards in images processed by Claude, is being introduced to meet the transparency obligations outlined in the European Union's AI Act. This regulation mandates that all synthetic audio, image, video, and text content must include machine-readable identifiers.

Why it matters

Professionals using or developing AI models need to understand evolving regulatory compliance, especially regarding content authenticity and transparency, to avoid legal issues and build trust.

How to implement this in your domain

  1. 1Stay informed about global AI regulations like the EU AI Act.
  2. 2Evaluate current AI content generation workflows for compliance gaps.
  3. 3Incorporate watermarking or content provenance tools into AI pipelines.
  4. 4Educate teams on the importance of AI transparency and ethical content creation.
  5. 5Develop internal policies for identifying and labeling AI-generated content.

Original post by AI | The Verge

"Anthropic has clarified how it's planning to apply invisible watermarks to Claude-generated text in order to comply with Europe's AI transparency rules. On Friday, Anthropic announced that Claude's text marking system is "a version of the SynthID-Text approach" - an open-source w…"

View on X

Originally posted by AI | The Verge on X · view source

Want to go deeper?

Turn these trends into skills with Learnijoy's hands-on AI & tech courses.

Explore courses