Anthropic Uses Google Tech to Watermark Claude Text
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Anthropic is applying invisible watermarks to Claude-generated text using a version of Google DeepMind's open-source SynthID-Text system to meet EU AI Act requirements.
- Anthropic says the watermarking uses low-stakes word choices in generated text to embed detectable patterns that are invisible to readers but require a secret key to identify.
- Anthropic confirms the watermarking will not affect the quality of Claude’s outputs or increase costs for users.
- Google DeepMind developed the SynthID-Text technology that Anthropic is adapting, and Google’s Gemini has supported this watermarking since 2024.
- EU AI Act mandates machine-readable labels for synthetic content, driving major AI developers like Anthropic and Google to implement text watermarking solutions.
Why it matters: AI companies must now bake compliance into core outputs without degrading performance—Anthropic’s use of subtle linguistic patterns shows a path forward that preserves user experience while meeting strict EU rules. This sets a precedent for how transparency can be enforced at scale without added cost or quality loss.
Ask SkimNews
