AI-designed bacteriophages overcome E. coli resistance in lab tests

Researchers have made significant strides in using bacteriophages—viruses that infect bacteria—to combat antibiotic resistance. Using an AI system named Evo, trained on millions of genetic sources, scientists at Stanford University and the Arc Institute successfully designed a cocktail of phages that overcame resistance in certain E. coli strains during laboratory tests. These AI-generated genomes differed from natural viruses, with some displaying evolutionary patterns not seen in nature.

Complementary research published in Science has identified how bacteria defend themselves against such viral attacks. Scientists at the University of Utah Health discovered that bacteria use a system called CBASS to detect specific enzymes used by invading viruses. When these enzymes cut sensor molecules within the bacterium, it triggers an immune response that can cause the infected cell to self-destruct, preventing the virus from spreading.

While these developments offer a potential path toward more effective phage therapies, they also raise biosecurity concerns. Experts at the Johns Hopkins Center for Health Security warned that the ability to compose viral genomes using generative AI has outpaced existing governance and safeguards, noting that research into eukaryote-infecting pathogens could pose significant risks.