on
Sixteen AI-designed viruses offer a new route against drug-resistant bacteria
In a world first, scientists led by a team from Stanford University have created 16 viable viruses that do not exist in nature and were designed by AI. Their experiment, which is published in Science, could help in the fight against superbugs by allowing researchers to design customized viruses to kill drug-resistant bacteria. Thomas Inglesby and Moritz S. Hanke have published a Perspective piece on the work and its implications in the same edition of the journal.
AI and medicine
Artificial intelligence is already helping to speed up drug discovery by analyzing massive genetic data sets, predicting protein shapes and identifying potential medicines. But in this research, the team wanted to see if AI could go a step further by creating an entire functioning genome based on a natural virus template.
How to build a virus
The scientists wanted AI to design novel bacteriophages, which are viruses that infect and destroy bacteria. So they first used specialized DNA language models that had already been trained on millions of DNA sequences. They then fine-tuned them on around 15,000 viral genomes from a family of small bacterial viruses known as Microviridae. This allowed the AI to learn the genetic rules governing functional viral genomes.
The software generated thousands of potential genome blueprints modeled after a standard phage called PhiX174. From these digital designs, the study authors synthesized 285 candidates into physical DNA sequences. They placed them inside host E. coli cells to see if they would form fully functional, reproducing viruses.
"We report the first generative design of complete bacteriophage genomes using genome language models," wrote the researchers in their paper.
Out of the 285 synthesized blueprints, 16 generated viable viruses that successfully infected and killed the target E. coli bacteria. In additional tests, a mixture of these synthetic viruses successfully killed bacterial strains that had developed resistance to the natural PhiX174 virus.
"Our results demonstrate that generative AI can capture an underlying evolutionary design space with enough fidelity to produce viable bacteriophage genomes."
Safety concerns
While these synthetic phages could one day be used to outsmart existing and evolving superbugs, researchers emphasize that clinical applications will require rigorous animal and human testing.
The team also took precautions by removing human-infecting viruses from the AI's training data. In their Perspective piece, Inglesby and Hanke echo these safety concerns. They also warn that biosecurity rules and oversight must catch up rapidly as AI genome tools evolve.
Written for you by our author Paul Arnold, edited by Lisa Lock, and fact-checked and reviewed by Robert Egan—this article is the result of careful human work. We rely on readers like you to keep independent science journalism alive. If this reporting matters to you, please consider a donation (especially monthly). You'll get an ad-free account as a thank-you.
Publication details
Samuel H. King et al, Generative design of bacteriophages with genome language models, Science (2026). DOI: 10.1126/science.aec2657
Thomas V. Inglesby et al, AI-designed viral genomes, Science (2026). DOI: 10.1126/science.aej8512
Who's behind this story?
BSc Biology from University of London. BBC documentary producer with world travel experience. Freelances from southern Spain. Full profile →
BA art history, MA material culture. Former museum editor, paramedic, and transplant coordinator. Editing for Science X since 2021. Full profile →
Bachelor's in mathematical biology, Master's in creative writing. Well-traveled with unique perspectives on science and language. Full profile →
© 2026 Science X Network
Citation: Sixteen AI-designed viruses offer a new route against drug-resistant bacteria (2026, August 10) retrieved 10 August 2026 from https://phys.org/news/2026-08-sixteen-ai-viruses-route-drug.html
This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no part may be reproduced without the written permission. The content is provided for information purposes only.