Scientists have used artificial intelligence (AI) to design complete viral genomes, producing 16 functional viruses capable of infecting bacteria, in a breakthrough that could help combat antibiotic resistance while raising concerns about biological safety.

The AI-designed viruses are bacteriophages, which infect bacteria rather than humans. Researchers say the technology could eventually help develop new treatments against antibiotic-resistant bacteria that no longer respond to conventional drugs.

The study, published in the journal Science on August 6, was conducted by researchers from Stanford University and the Arc Institute using a generative AI model called Evo 2.

The researchers generated hundreds of potential viral genomes based on an existing bacteriophage and synthesised nearly 300 designs for laboratory testing. Sixteen successfully produced functional viruses capable of infecting E. coli bacteria. A combination of the AI-designed viruses also overcame resistance in bacterial strains that had become resistant to the original virus. Stanford Report+1

However, the breakthrough has raised biosecurity concerns among scientists. In a related commentary published alongside the study, researchers from Johns Hopkins University warned that the ability to generate functional viral genomes carries significant biosafety and biosecurity implications. PubMed+1

Scientists stress that the experiment did not produce viruses capable of infecting humans. However, the technology’s potential misuse in designing harmful biological agents has prompted calls for careful oversight as AI capabilities in biotechnology continue to advance.