Researchers at the Wyss Institute for Biologically Inspired Engineering, Harvard University, and MIT have demonstrated an artificial intelligence-driven pipeline capable of identifying highly selective antibacterial candidates against multidrug-resistant gonorrhea. The study, published in Science Translational Medicine, outlines a deep learning methodology that screened six million virtual compounds to isolate two lead molecules with potent activity against Neisseria gonorrhoeae, the pathogen responsible for tens of millions of sexually transmitted infections globally. The development addresses a critical public health bottleneck. Despite the recent approval of zoliflodacin and gepotidacin as the first entirely new antibiotic classes for uncomplicated urogenital gonorrhea in over three decades, epidemiological models predict inevitable resistance within a decade of widespread deployment. To circumvent this predictable cycle, the research team, led by Wyss Core Faculty member James Collins and spearheaded by clinician-scientist Melis Anahtar, engineered a predictive machine learning framework trained on a dataset of 38,650 experimentally validated small molecules. The model was optimized to recognize chemical architectures that diverge from conventional antibiotics, thereby targeting unconventional bacterial pathways to delay resistance development. Applying the algorithm to an ultra-large virtual library yielded 213 putative hits, which underwent rigorous growth-inhibitory, resistance, and cytotoxicity screening. This cascade ultimately narrowed the candidate pool to two structurally distinct compounds. Proteomic analysis identified compound A1, an aminothiazole derivative that specifically inhibits alanine racemase, a critical enzyme required for bacterial cell wall synthesis. This mechanism represents a novel therapeutic approach for gonorrhea, distinct from existing cell-wall inhibitors, and demonstrated exceptionally low resistance induction rates in vitro. The second candidate, MP20, was evaluated using a human vaginal organ-on-a-chip developed by Wyss Founding Director Donald Ingber laboratory, where it significantly reduced pathogen titers upon exposure to vaginal epithelial cells. Translational validation followed in a murine model of vaginal gonococcal infection. Administration of compound A1 across five doses within a 24-hour window substantially decreased bacterial load compared to untreated controls. While both molecules remain in the preclinical hit-to-lead phase and require further medicinal chemistry optimization, the study establishes a scalable discovery architecture. By integrating high-fidelity biological datasets with generative artificial intelligence, the pipeline can rapidly interrogate make-on-demand chemical libraries, identifying therapeutically viable compounds that traditional screening methods would likely overlook. This convergence of computational biology and human-relevant physiological models underscores a paradigm shift in antimicrobial research, offering a durable strategy to maintain efficacy as N. gonorrhoeae continues to evolve.