Anthropic explains how Claude’s invisible text watermarks will work
Anthropic is embedding invisible fingerprints into every Claude response — here's how it works and why it matters for compliance.
Why it matters
Anthropic is implementing watermarking in Claude to comply with EU AI Act transparency mandates. This is a significant regulatory compliance milestone that signals how frontier labs are operationalizing AI governance in practice.
The key facts
11 to knowAnthropic adopting Google DeepMind's open-source SynthID-Text watermarking technology
Watermarks create detectable patterns using wording probabilities
Feature required to meet EU AI Act obligations for synthetic text disclosure
Alongside C2PA support for Claude-processed images
EU AI Act mandates machine-readable marks on synthetic audio, image, video, and text
Anthropic implementing SynthID-Text watermarking (open-source from Google DeepMind)
System uses wording probability patterns to create detectable marks in text
Feature introduced to comply with EU AI Act transparency requirements
Also rolling out C2PA support for Claude-processed images
EU AI Act mandates machine-readable marks for synthetic audio, image, video, and text
Announcement published August 17, 2026
Go to the source
The Verge AItheverge.com
Publisher excerpt: 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 watermarking…