Updated
Updated · University of Minnesota Twin Cities · Aug 17
AI-Designed Phages Kill 2 Resistant E. coli Strains as Experts Urge Genomics Safeguards
Updated
Updated · University of Minnesota Twin Cities · Aug 17

AI-Designed Phages Kill 2 Resistant E. coli Strains as Experts Urge Genomics Safeguards

3 articles · Updated · University of Minnesota Twin Cities · Aug 17

Summary

  • 16 complete bacteriophages designed by Stanford and Broad Institute researchers included candidates that rapidly destroyed E. coli strains resistant to naturally occurring ΦX174-like phages, according to a Science study.
  • Using genomic language models and computational biology, the team generated whole viral genomes with substantial differences from the original phage; some engineered versions performed comparably to natural relatives.
  • Johns Hopkins commentators said the work opens a path to adaptive phage therapies against fast-evolving pathogens but exposes a governance gap because generative AI can now compose viral genomes.
  • They urged adapting biosafety and biosecurity frameworks for generative genomics and excluding sensitive viral sequences from training data to reduce risks of designing eukaryote-infecting pathogens.

Insights

As governments rush to deregulate synthetic biology, will you ever truly know if your medicine was invented by a machine?
If an AI language model hallucinates text, what happens when it hallucinates the genetic code of a live virus?