The first antibiotic fully designed by AI enters Phase I trials, showing potent activity against MRSA and other drug-resistant superbugs.
The first antibiotic designed entirely by artificial intelligence has entered Phase I clinical trials, marking a potentially transformative moment in the battle against antimicrobial resistance. The compound, designated ABX-1701, was discovered and optimized by an AI platform developed by researchers at MIT and the Broad Institute, and it demonstrates potent activity against several of the most dangerous drug-resistant bacteria.
Antimicrobial resistance (AMR) is one of the greatest threats to global health. The World Health Organization estimates that drug-resistant infections directly caused 1.27 million deaths in 2019, a figure projected to reach 10 million annually by 2050 if current trends continue. Despite this escalating crisis, antibiotic development has stagnated, with no truly novel class of antibiotics reaching the market since the 1980s.
The AI platform, called Synthia, was trained on a database of over 100 million chemical compounds and their biological activities. Using deep learning and generative chemistry models, Synthia identified structural motifs associated with antibacterial activity and designed novel molecules that target bacterial membranes through a mechanism distinct from any existing antibiotic class.
In preclinical testing, ABX-1701 showed efficacy against methicillin-resistant Staphylococcus aureus (MRSA), vancomycin-resistant Enterococci (VRE), and carbapenem-resistant Acinetobacter baumannii, all classified as critical priority pathogens by the WHO. Importantly, laboratory studies suggest that bacteria develop resistance to ABX-1701 at a rate approximately 100 times slower than to existing antibiotics.
Dr. James Collins of MIT, who co-led the research, explained that the AI approach allows exploration of chemical space far beyond what human chemists could achieve. The platform evaluated billions of potential molecular structures in weeks, a task that would take conventional drug discovery programs decades.
The Phase I trial, enrolling 60 healthy volunteers, will assess the safety, tolerability, and pharmacokinetics of ABX-1701. If successful, Phase II trials in patients with drug-resistant infections are planned to begin in early 2026.
The project was funded by a combination of government grants, philanthropy, and a novel subscription model in which participating healthcare systems pay an annual fee for access to the antibiotic if approved, ensuring commercial viability regardless of the volume of prescriptions. This model addresses the perverse economic incentive that has historically discouraged antibiotic development.