AI-designed bacteriophages kill E. coli in lab

16 Oct 2025

Stanford researchers used an AI called Evo to design bacteriophages from genomic data and produced real, working viruses that infect and kill E. coli. The team trained Evo on roughly 2 million bacteriophage genomes, asked it to propose new viral sequences, chemically synthesized 302 of those AI-designed genomes, and tested them in the lab. Sixteen designs successfully replicated in E. coli and produced plaques — visible zones of killed bacteria — demonstrating that AI-generated genomes can produce functional phage particles.

The work matters because antibiotic resistance is rapidly growing: the WHO recently reported that about 1 in 6 bacterial infections worldwide are now resistant to antibiotics, over 40% of common antibiotics showed increased resistance between 2018 and 2023, and in some regions more than 70% of E. coli resist first-line treatments. Phage therapy (using viruses that attack bacteria) is already a clinical option for some drug-resistant infections; AI-designed phages could speed development and broaden available therapies, testing hundreds of candidates far faster than traditional methods.

The researchers deliberately limited the model (for example, not teaching it about human viruses) and note safety concerns — the same capability could be misused if applied to dangerous human pathogens. Parallel investments such as purpose-built, automated AI robotic labs (Liverpool’s new £20M facility is cited) aim to accelerate safe development of antiviral and antibacterial treatments while highlighting the need for strong safeguards.

Source