AI designs first functional viruses not found in nature

Researchers at Stanford University and the Arc Institute have used artificial intelligence to design 16 functional viruses with genomes never before seen in nature, in what scientists describe as a "very significant turning point" for synthetic biology. The work, published Thursday in the journal Science, marks the first time generative AI has been used to create complete, viable viral genomes.bbc+1

How It Works

The team used AI models called Evo 1 and Evo 2, which function similarly to large language models like ChatGPT but predict genetic sequences rather than words. Trained on genetic data from viruses, bacteria, plants, and humans, the models were fine-tuned to design bacteriophages — viruses that target bacteria — modeled on a naturally occurring virus called Phi X-174 that infects only E. coli.biorxiv+1

From approximately 300 AI-generated genome designs synthesized in the laboratory, 16 proved viable, successfully infecting and killing E. coli bacteria. Several of the AI-designed phages demonstrated higher fitness than the natural template virus and were able to overcome bacterial resistance that the original phage could not.nytimes+2

"This is the first time generative AI has been employed to devise a complete genome that can replicate and perform other cellular functions," Brian Hie, an assistant professor at Stanford University, told the BBC.bbc

Medical Promise

The breakthrough opens a path toward new treatments for antibiotic-resistant infections. A cocktail of the AI-generated phages rapidly overcame resistance in three E. coli strains during testing, while the natural virus alone failed completely. Phage therapy is increasingly viewed as a promising approach to combat bacterial infections that no longer respond to traditional antibiotics.arcinstitute+2

Patrick Cai, chair of genomics at the Manchester Institute of Biotechnology, called the study an "important milestone," noting that genome language models "are beginning to grasp the design principles encoded by evolution."bbc+1

Biosecurity Concerns

In a commentary accompanying the Science publication, Thomas Inglesby and colleagues at the Johns Hopkins Center for Health Security warned the findings raise "urgent safety and security questions," stating it is no longer a matter of whether generative viral design will arrive but how it can proceed "without enabling serious harm."bbc

The researchers took precautions by excluding viruses capable of infecting complex organisms from their training data and working exclusively on bacterial phages within a secure laboratory environment. The engineered phage genomes comprise about 5,400 base pairs — a fraction of the roughly 500,000 base pairs in the smallest living cell's genome — but Hie indicated his team is "definitely interested in working towards" designing more complex organisms.bbc