Evo AI Models Design 16 New Viruses to Kill Bacteria

Scientists at Stanford University and the Arc Institute have successfully used artificial intelligence to design 16 completely new, functional viruses from scratch. Published in the journal Science, the study leveraged two foundational AI models for computational biology, Evo 1 and Evo 2. These algorithms, trained on millions of genomes across all domains of life, analyzed evolutionary patterns to generate synthetic genomes. Rather than copying existing pathogens, the AI created novel genetic sequences that are functional but distinct from any found in nature.

The research team focused on bacteriophages, which are simple viruses that target bacteria. Using the Phi X-174 bacteriophage—a virus that infects Escherichia coli—as a structural guide, the AI generated thousands of potential genomes. From this pool, researchers selected 300 promising designs based on gene organization and regulatory elements. They then chemically synthesized these 300 genomes piece by piece in a laboratory. When introduced to E. coli, 16 of the synthesized genomes successfully produced fully functional, replicating viruses with unique gene sequences and varying genome sizes.

For biotechnology practitioners, this represents a major shift toward automated, programmable medicine. In laboratory tests, the AI-generated phages successfully bypassed bacterial defenses, rapidly infecting E. coli strains that had already developed resistance to the natural Phi X-174 virus. This capability suggests that AI can be used to rapidly engineer custom therapies that evolve alongside mutating pathogens, offering a powerful new weapon against antibiotic resistance.

However, the ability to generate viable, novel pathogens also introduces severe biosecurity risks. Experts warn that the technology could be misused to engineer dangerous biological weapons or new diseases. Moritz Hanke of the Johns Hopkins Center for Health Security noted to The New York Times that a "huge disconnect" exists between rapid scientific progress and the development of regulatory frameworks, highlighting the urgent need for safeguards as AI-driven synthetic biology matures.