AI designs working viruses from scratch; researchers flag biosecurity risks

The work by Samuel King and colleagues represents a significant leap in computational biology: the ability to design complete, functional genomic systems from first principles rather than tweaking existing sequences. Previous advances in DNA sequencing and synthesis made it increasingly possible to read and write entire genomes, but designing a functional genome from scratch remained extraordinarily difficult because genes, regulatory sequences, and other elements interact in highly complex ways. King's team bridged this gap by combining Evo genomic language models (including Evo 1), computational biology, and experimental screening—a methodological integration that allowed them to generate and test hundreds of candidate designs efficiently.

The practical demonstration using bacteriophages—viruses that infect bacteria—grounds the work in a domain with immediate therapeutic relevance. Phage therapy has long been considered a potential tool against antibiotic-resistant bacteria, but designing phages that remain effective as bacteria evolve resistance has been a persistent challenge. The finding that some engineered phage combinations could overcome resistance in E. coli strains that had become resistant to ΦX174-like phages suggests this AI-guided approach could unlock a new generation of durable therapeutics.

At the same time, commentators Thomas Inglesby and Moritz Hanke note that the capability to design and synthesize functional AI-generated viral genomes introduces a novel and serious risk: the same generative capacity that enables therapeutic design could potentially be misused to engineer harmful pathogens. The researchers acknowledge this explicitly and call for robust oversight, including consulting safety and security professionals throughout the design process and adapting existing safety frameworks to this new domain. Their framing—that the question is no longer whether generative viral genome design will exist, but whether society can build oversight that allows its benefits while preventing harm—reflects recognition that the technology is now inevitable and that governance structures must evolve alongside it.