AI Creates 16 New Viruses That Fight Resistant Bacteria

For the first time, artificial intelligence has successfully designed a series of previously unknown viruses capable of infecting and eliminating certain types of bacteria. Researchers at Stanford University and the Arc Institute used AI models called Evo 1 and Evo 2 to achieve this breakthrough. Both algorithms were trained on millions of genomes from all domains of life—animals, plants, microbes, bacteria, and viruses found in nature—to learn complex evolutionary patterns, including how genes are organized, which sequences are conserved, and the biological constraints that allow organisms to function.

The team focused on bacteriophages, viruses that infect only bacteria and have relatively small genomes that are easy to synthesize and manipulate in controlled laboratory conditions. The experiment used the bacteriophage Phi X-174, which naturally infects the bacterium E. coli, as a functional reference guide. The goal was not to reproduce this virus but to use it as a template to teach the AI algorithms to generate thousands of entirely new genomes capable of infecting E. coli. The key insight was that the AI-generated viruses retained the functional organization essential for recognizing bacteria, inserting their DNA, replicating it, and assembling new viral particles—but the specific DNA sequences differed considerably from naturally occurring phages.

Scientists then evaluated the AI-generated genomes, considering gene organization, regulatory elements, and other biological criteria inspired by Phi X-174. This resulted in a selection of 300 candidate genomes, which were artificially synthesized molecule by molecule in the laboratory and introduced into E. coli bacteria. Of the 300 synthesized genomes, only 16 produced fully functional bacteriophages with previously unpublished sequences, different genes, new regulatory elements, and varying genome sizes. Notably, the behavior of these viruses also varied: some infected bacteria more quickly, while others exhibited different abilities to replicate.

In a critical test, researchers exposed a mixture of AI-designed phages and a mixture of natural phages similar to Phi X-174 to E. coli strains that had already developed resistance to the natural virus. The AI-generated viruses rapidly overcame bacterial resistance and established infection. According to the authors, this finding demonstrates "a path toward artificial intelligence–generated phage therapies against rapidly evolving bacterial pathogens." The potential clinical application is significant: personalized treatments could be developed that evolve at nearly the same rate as the pathogens themselves.

The research, published this week in the journal Science, has also triggered a serious conversation about biosecurity. Moritz Hanke, a researcher at Johns Hopkins Center for Health Security, told The New York Times that there are currently no safeguards capable of effectively preventing the creation of a lethal virus with AI assistance, describing a "huge disconnect" between the speed of scientific advancement and the development of effective regulatory frameworks. The concern is not entirely new—three years ago, a Rand Corporation study warned that the most advanced AI systems had the capacity to refine the planning and execution of attacks using biological weapons. With the rapid development of AI technology, experts warn that such capabilities will likely become even greater and more sophisticated, and that the speed at which AI systems evolve often outpaces governments' capacity for regulatory oversight.