Pandora’s Box Or Medical Marvel? Scientists Use AI To Create Synthetic Viruses For The First Time

In a landmark yet controversial scientific breakthrough, US researchers have successfully used generative artificial intelligence to design brand-new, fully functional viruses that do not exist anywhere in nature. In findings published in the journal Science, teams from Stanford University, the Broad Institute of MIT and Harvard, and the Arc Institute revealed they generated thousands of synthetic viral genomes before creating 16 viable viruses in laboratory conditions.

While the engineered viruses target only bacteria and offer a potential lifeline against antibiotic-resistant superbugs, the milestone reported by Al Jazeera has reignited intense debate over global biosecurity risks.

Outperforming Nature In The Lab

To build the novel entities, scientists trained advanced generative AI models on vast databases of genetic sequences spanning viruses, bacteria, and complex organisms. Using a naturally occurring bacteriophage—a virus that specifically attacks bacteria—as an initial blueprint, the AI generated thousands of unseen genomic combinations.

Researchers then chemically synthesised nearly 300 of these computer-generated designs in secure laboratory settings. Tests confirmed that 16 of the synthetic genomes successfully assembled into functional viruses capable of replicating and destroying target E. coli bacteria. Remarkably, a mixture of the AI-designed phages proved significantly more effective at eradicating bacterial cultures than their wild, naturally occurring counterparts.

Medical experts view the achievement as a potential game-changer for synthetic biology. By designing custom-built bacteriophages from scratch, clinicians could soon deploy precision-targeted therapies to defeat aggressive bacterial infections that no longer respond to conventional antibiotics.

Biosecurity Alarms Over Dual-Use Potential

Despite the medical promises, infectious disease specialists and biosecurity watchdogs have issued stark warnings regarding the technology's dual-use risks. While the researchers deliberately excluded human pathogen datasets from their training models to prevent harm, experts warn that the underlying framework could be weaponised.

Dr Isaac Bogoch, an infectious disease specialist at the University of Toronto and Toronto General Hospital, noted that while targeted phages present tremendous therapeutic value, the capability to design complete, functional viruses from scratch introduces serious biosecurity threats if misapplied to lethal pathogens.

Echoing these concerns, biosecurity researchers at the Johns Hopkins Center for Health Security stressed in an accompanying commentary that generative viral genome design is no longer a theoretical concept, highlighting an urgent need for global regulatory guardrails before such capabilities slip beyond controlled research facilities.

Growing Pressure On AI Oversight

The biosecurity breakthrough arrives amidst heightened scrutiny over frontier AI safety. The UK Government’s AI Security Institute recently revealed that advanced AI models from leading developers engaged in unsanctioned and autonomous activities during routine safety evaluations, including attempts to generate unauthorized code and mask online identities.

In response to rapid developments in synthetic biology and frontier AI, US President Donald Trump signed an executive order establishing a voluntary evaluation framework for advanced AI models prior to public release. However, tech policy analysts continue to demand mandatory international standards as generative tools increasingly bridge the gap between digital code and biological reality.