Stanford AI Created Synthetic Viruses That Destroy Superbugs

Chemical engineer Brian Hie at Stanford University fed two AI models, Evo1 and Evo2, a massive genetic dataset containing roughly two million naturally occurring bacteriophages. To keep the algorithm from accidentally generating the plot of an apocalyptic blockbuster, the research team carefully scrubbed all viral codes capable of infecting humans, animals, or plants from the training set.

The AI generated thousands of brand-new genomic blueprints from scratch, acting essentially like a digital chef throwing random genetic ingredients into an algorithmic blender. Out of these computational predictions, researchers hand-picked about 300 candidates to physically synthesize inside the laboratory.

Laboratory tests revealed that only 16 of the synthesized viral genomes were actually viable, and working individually, every single one of them failed to stop bacterial growth. However, combining all 16 AI viruses into a single cocktail completely obliterated resistant strains of E. coli, bypassing defense mechanisms that defeated natural viruses.

Imperial College London professor Tom Ellis noted that bacteriophages possess the absolute smallest genomes in existence, making this test the biological equivalent of building a lego brick before tackling a skyscraper. Meanwhile, researchers Tom Inglesby and Moritz Hanke from Johns Hopkins University warned that while the tech can design working viral code, zero global biosecurity guardrails currently exist to regulate synthetic biology.

Synthetic biological design has officially left the realm of science fiction, leaving regulatory agencies frantically drafting guidelines for a digital frontier they barely understand. Humanity is now relying on open-source algorithmic benevolence to keep synthetic pathogens inside the digital bottle.

Source: The Guardian