Google DeepMind introduced SynthID Bio, a family of watermarking methods for synthetic biology that embeds an imperceptible, verifiable signature into AI-generated biological code — protein sequences and predicted 3D structures — while preserving biological function. For protein folding it fine-tunes a small part of AlphaFold 3's diffusion network so the signature sits in the model weights; for binders it used AlphaProteo alongside a SynthID Bio-enabled version of ProteinMPNN. In wet-lab testing across three target proteins (VEGF-A, the SARS-CoV-2 spike RBD and PD-L1), the watermarked designs matched unwatermarked versions on hit rate, binding affinity and natural sequence diversity. DeepMind frames it as a verification layer for DNA-synthesis screening and for databases such as the Protein Data Bank, UniProt and GenBank.
Our takeWatermarking text was always contestable; watermarking a physical molecule is a harder, cleaner test — either the signature survives synthesis or it does not. The deeper signal is the target: biofoundries and sequence databases, where provenance matters more than in a chat log. If it holds, whether a protein was designed by AI becomes answerable at the bench.