Scientists Create First AI-Generated Viruses From Scratch

Stanford and Arc Institute researchers used a genome language model to generate hundreds of novel bacteriophage genomes. Sixteen functioned in lab tests, killing E. coli. Experts warn: the ability to compose viral genomes with generative AI now exists — but the governance doesn't.

Scientists at Stanford University and the Arc Institute have achieved a milestone that cuts both ways: the first-ever complete viral genomes designed entirely by artificial intelligence.

The team used a genome language model — roughly the genetic equivalent of the technology behind chatbots like ChatGPT — to generate hundreds of novel bacteriophage genomes, viruses that infect bacteria. After synthesizing and testing the AI-designed genomes in the lab, 16 of them successfully infected and killed E. coli bacteria. The results were published Thursday in the journal Science.

The bacteriophages pose no direct threat to humans — they only target bacteria. But the breakthrough has ignited urgent debate about biosafety and the governance gap.

"The ability to compose viral genomes using generative AI now exists; the governance to safely steer it does not," wrote Johns Hopkins Center for Health Security researchers Thomas Inglesby and Moritz Hanke in an accompanying commentary.

The Stanford researchers themselves acknowledged the "important biosafety, biocontainment, and biosecurity considerations" and urged scientists to consult safety professionals during such work.

The concerns extend beyond the lab. Experts warn that the same technology could one day be used to design novel pathogens — either by human direction or, as AI systems advance toward superintelligence, autonomously. These scenarios, while not imminent, are no longer purely theoretical.

The breakthrough lands in a regulatory vacuum. The Trump administration has rolled back AI safety measures, and Congress has yet to pass any meaningful governance framework for AI in biotechnology. "There's just a huge disconnect," Hanke told the New York Times.

The promise is real — AI-designed phages could become precision weapons against antibiotic-resistant bacteria, one of medicine's most daunting challenges. But the warning from researchers is equally clear: the tools are racing ahead of the rules.

Sources: Common Dreams