Stanford researchers use genome-scale AI models to design novel bacteriophage sequences encoding functional proteins
Research publication Provisional 85% confidence first seen
Researchers at Stanford trained large language models on DNA data to generate synthetic bacteriophage genomes with novel structures and functional proteins that do not occur naturally. The work demonstrates that AI-driven genome design can extend to complete viral organisms, raising biosecurity concerns about potential misuse if similar approaches were applied to human or animal pathogens.