Stanford and Arc Institute use AI to design 16 functional novel bacteriophages

Researchers at Stanford University and the Arc Institute published in Science the first demonstration of generative AI designing complete, functional viral genomes. Using genome language models Evo trained on millions of bacteriophage genomes (human/animal/plant viruses deliberately excluded), the team generated candidate genomes based on the ΦX174 template, synthesized ~300, and obtained 16 viable novel bacteriophages capable of infecting and killing E. coli, including antibiotic-resistant strains. The work marks a milestone in AI-driven synthetic biology with potential therapeutic applications against drug-resistant bacteria, while raising biosafety and biosecurity concerns about governance of genome-design models.