AI Designs 16 Brand-New Viruses From Scratch — A Scientific First

Stanford researchers used AI models Evo1 and Evo2 to design complete viral genomes from scratch, producing 16 fully functional bacteriophages that kill E. coli bacteria. The breakthrough, published in Science, opens new paths for combating antibiotic resistance but raises urgent biosafety concerns.

Stanford researchers have used artificial intelligence to design 16 brand-new viruses from scratch — the first time AI has successfully designed complete, functional viral genomes that can replicate in the laboratory.

The breakthrough, published in the journal Science, marks what experts are calling a "very significant turning point" for synthetic biology. The AI models, named Evo1 and Evo2, were trained on genetic codes from viruses, bacteria, plants, and humans — then refined to generate bacteriophages, viruses that infect and kill specific species of bacteria.

Out of 302 AI-designed genomes synthesized in the lab, 16 proved fully functional against E. coli bacteria. Samuel King, a PhD student in the Stanford lab, described the moment they saw clear spots forming on the petri dishes: signs the phage were successfully killing bacteria. "The room spontaneously burst into applause," recalled Brian Hie, the assistant professor who led the work.

The immediate promise is phage therapy — using engineered viruses to combat the growing crisis of antibiotic-resistant infections, which already kill over a million people annually. Unlike broad-spectrum antibiotics, phage can be precisely targeted to specific bacterial strains.

But the findings also sharpen an urgent biosafety debate. Researchers from the Johns Hopkins Center for Health Security, in a commentary accompanying the paper, warned the work raises "urgent biosafety and biosecurity questions," arguing it is no longer a matter of whether generative viral genome design will exist, but whether it can be deployed "without enabling serious harm."

The researchers took precautions: they excluded viruses capable of infecting complex organisms from the training data, worked exclusively on phage that target bacteria, and conducted all work in a secure laboratory. Hie argues existing safeguards "go a long way towards ensuring that the technology is used for good."

The phage genomes are relatively small at around 5,400 base pairs — a fraction of the 3 billion in the human genome. But the field is advancing fast. Hie said attempting simple organisms was "definitely interesting" and "probably a lot of work, but not impossible."

Sources: BBC News, WIRED