Google Develops Watermarking Technique for AI-Designed Proteins

Google has announced a new method to watermark proteins generated by artificial intelligence, a development that could help identify and track AI‑designed molecules. The move comes amid growing concerns that AI tools, while enabling breakthroughs such as enzymes that break down plastics or neutralize venom, also pose biosecurity risks by making it easier to create toxins or modify viral proteins. Current software used to flag potentially dangerous DNA sequences fails to recognize AI‑generated proteins because they lack established threat signatures. By embedding a watermark, Google aims to give regulators and researchers a way to detect AI‑derived proteins and assess their safety before they are deployed.