AI Uncovers Entirely New Type of Antibiotic to Fight Drug-Resistant Bacteria
An AI program designed to find new medicines that kill bacteria screened billions of compounds. It discovered a completely new kind of chemical that works against tough, drug-resistant bacteria (often called superbugs). This new chemical belongs to a family of molecules that scientists never before thought could be antibiotics.
This breakthrough offers a fresh weapon in the urgent global fight against increasingly dangerous drug-resistant infections.
Antibiotic discovery has stalled for decades because traditional screening rarely yields chemically novel hits. A research group trained neural networks to predict whether molecules would inhibit the growth of a resistant pathogen, then applied the model to libraries far larger than any physical screen could test.
The model prioritized a compound with an unusual scaffold. Follow-up experiments confirmed activity against resistant strains while sparing helpful gut bacteria, and showed low toxicity in early animal tests.
The breakthrough is less about any single molecule than about the search itself: AI explored a chemical space far beyond human intuition and returned a genuinely new starting point for desperately needed antibiotics.
🔬 Read the technical version
Machine learning screens billions of compounds to surface a new structural class of antibiotic
A deep-learning model trained to predict antibacterial activity flagged a chemically novel compound effective against drug-resistant bacteria — a structural class chemists had not previously associated with antibiotic action.