US scientists released new research on Thursday outlining how, in a scientific first, they used generative artificial intelligence to design new viruses that are not found in nature, potentially paving the way for the creation of entire biological systems at the genome scale.

The findings, published in the journal Science, said the researchers, “used generative AI models trained on millions of natural genomes to design entire bacteriophages,” which are viruses that infect and can replicate within bacteria.

The scientists from Stanford University and the Arc Institute, a California-based non-profit dedicated to “high-risk, high-reward” research, used two “genome language models” named Evo 1 and Evo 2 in their design process.

Evo 1 and 2 work on principles similar to those of large language models, but instead of writing text they predict genetic sequences. Their training data consisted of the genomes of viruses, bacteria, and more complex organisms such as plants and animals.

The researchers chemically synthesized nearly 300 genomes and tested them in laboratory conditions. The experiments produced 16 artificially engineered bacteriophages with “diverse sequences, structures, and fitness profiles.”

When put to the test, the engineered viruses proved more effective at killing E coli bacteria than the naturally occurring viruses.

“Our approach expands what synthetic genomics can achieve alongside methods such as directed evolution and rational engineering, lays out a path for generating adaptive and resilient phage therapies against rapidly evolving pathogens, and establishes a foundation for the generative design of larger, more complex genomes,” the study wrote.

The researchers said this first step into AI-based genome design could pave the way for a broader toolkit of genome sequencing and synthesis for larger biological systems and treatments.​​​​​​​