ProteinMPNN (AI model): news timeline & CVEs

Google figures out how to watermark AI-designed proteins

Google DeepMind's SynthIDBio watermarks AI-designed proteins without breaking function, aiding biosecurity checks.

Google DeepMind published SynthIDBio, which adapts its SynthID watermarking system to proteins designed with ProteinMPNN. A key biases amino-acid choices only when they remain compatible with a functional protein, spreading a detectable signal along the chain. Existing screens often miss novel AI-designed proteins, so marked sequences from trusted researchers can be identified and unmarked designs scrutinized. The work responds to concerns that the same tools used for beneficial enzymes could also design toxins or alter viral proteins.

Ars Technica · AIupdated 03:31 UTC · 58m agofirst Sep 30, 15:54 UTC · 12h agoAI safety & security 3 sources

Introducing SynthID Bio

Google DeepMind unveiled SynthID Bio, a watermark for AI-designed proteins that remains detectable after synthesis.

Google DeepMind introduced SynthID Bio, watermarking methods that place a detectable signature in AI-generated protein sequences and predicted 3D structures while aiming to preserve function. In wet-lab tests, watermarked protein binders for VEGF-A, the SARS-CoV-2 spike RBD, and PD-L1 matched unwatermarked hit rate, binding affinity, and sequence diversity using AlphaProteo and ProteinMPNN. For folding, a fine-tuned part of AlphaFold 3 embeds the mark in predicted coordinates with reported near-perfect detection and preserved accuracy. DeepMind and Twist Bioscience describe it as a provenance signal for DNA synthesis screening and public sequence databases.

Google DeepMindupdated 03:31 UTC · 58m agofirst Sep 30, 15:03 UTC · 13h agoAI safety & security 3 sources

ProteinMPNN

Timeline2 articles

Appears with

Google figures out how to watermark AI-designed proteins

Introducing SynthID Bio