Researchers at Stanford University and the Arc Institute in Palo Alto trained an A.I. model, called Evo, to learn the “grammar” of DNA. They then asked it to write new viral genomes. Out of around 285 A.I.-generated DNA sequences they tested, 16 produced real, living viruses in a Petri dish. These viruses infect bacteria, not humans, but the experiment shows that A.I. can design biological organisms from scratch.
This represents a huge shift in how biological research might be done in the future. For years, scientists have been able to copy or modify viruses. However, creating entirely new ones has always required deep knowledge of genetics. Now, an A.I. model trained on billions of DNA letters can generate viral blueprints in seconds. Supporters say this could lead to new medical tools, like better gene‑therapy delivery systems. Critics also warn that the same technology could be misused to create dangerous pathogens. This study demonstrates how the debate over A.I. and biotechnology is becoming crucial.
