Google DeepMind announced a new tool called SynthID Bio on September 30, 2026. The system embeds an invisible watermark directly into the amino‑acid sequence or 3‑D structure of AI‑generated proteins. The watermark can be detected later, even after the protein is synthesized in the lab, and it does not change the protein’s biological activity.
How the watermark works
SynthID Bio nudges the AI model to pick certain amino acids or tweak atomic coordinates just enough to create a detectable pattern. In tests with three targets – VEGF‑A, the SARS‑CoV‑2 spike RBD, and PD‑L1 – the watermarked proteins bound just as strongly as the non‑watermarked versions. The same approach was used to embed a signature into AlphaFold 3’s predictions, so the 3‑D coordinates themselves carry the mark.
Why it matters
The watermark gives DNA‑synthesis providers a way to flag orders that came from trusted AI models, reducing manual review time. It also helps public databases spot mislabeled AI‑generated entries, protecting downstream research from accidental biosecurity risks.
Why it matters
For researchers, the watermark means they can prove a protein design came from a trusted AI system without compromising function. For biosecurity teams, it offers an automated check that can speed up DNA‑synthesis screening and keep public protein databases cleaner.
