Antimicrobial resistance has become a major concern in recent years. According to studies, approximately 1.1 million deaths each year are linked to bacterial resistance to antimicrobial drugs.[1] The fear of antibiotic resistance continues to grow, as explained by recent reports from the U.S. Centers for Disease Control and Prevention showing that rates of dangerous bacterial infections surged by 69% between 2019 and 2023, with certain so-called “nightmare bacteria” being especially difficult to treat using existing antibiotics.[2] There are also concerns that when antibiotics kill harmful bacteria they kill beneficial gut bacteria as well.[3] Researchers have long sought narrow-spectrum antibiotics—drugs that selectively kill harmful bacteria while sparing beneficial species—but such compounds have been notoriously difficult to identify and validate.[4]
Artificial intelligence (“AI”) has rapidly emerged as a transformative tool across scientific and medical applications, becoming an asset in developing new antibiotics. Generative AI can design peptides (short amino acid chains) that have antimicrobial properties and can destroy pathogens, as well as rank them based on how effective they might be at killing various types of bacteria.[5] This allows researchers to create possible antibiotics and test them within the span of weeks, a process that once required many years of laboratory screening and testing.[6] AI allows researchers to sift through these potential drug molecules faster and more efficiently.
Generative AI modeling also allows researchers to discover the mechanisms of how antibiotics work and how they bind in the body. Determining a drug’s mechanism of action normally requires years of experimentation, but AI can speed up this process.[7] Additionally, many drugs are discovered by phenotypic screening, a process in which researchers observe the effects of a drug without knowing the underlying drug-protein interactions.[8] Thus, discovering a drug’s mechanism of action can be critical for researchers figuring out drug improvements, as well as identifying potential side effects, especially before clinical trials take place.[9] This process, known as “reverse screening,” thereby may reduce adverse side effects and help researchers narrowly target harmful bacteria.[10]
A notable example comes from researchers at MIT who used generative AI to identify enterololin, a novel compound that suppresses bacteria linked to Crohn’s disease flare-ups while leaving the rest of the microbiome largely intact.[11] In using AI to discover this more targeted approach to Crohn’s treatment, researchers were able to map how the compound worked in a span of months instead of years.[12] Mechanistic studies that once required 18–24 months were completed in roughly six months at a fraction of the cost required using a traditional approach.[13] Precision antibiotics, like enterololin, may also help alleviate antimicrobial resistance in the gut, as the compound acts on specific Crohn’s disease targets while leaving beneficial bacteria intact.[14] MIT’s use of AI to develop a potential treatment for Crohn’s disease is just the start of a new wave of drug discovery with potentially sweeping benefits.
The integration of AI into antibiotic discovery marks the beginning of a new era in precision microbiology—one where computational design, mechanistic insight, and clinical translation converge to address one of medicine’s most formidable challenges.
Editor: Brenden S. Gingrich, Ph.D.
[1] See Rachel Fieldhouse, AI has designed thousands of potential antibiotics. Will any work?, Nature (Oct. 3, 2025), https://www.nature.com/articles/d41586-025-03201-6.
[2] See id.
[3] See Rachel Gordon, AI maps how a new antibiotic targets gut bacteria, MIT News (Oct. 3, 2025), https://news.mit.edu/2025/ai-maps-how-new-antibiotic-targets-gut-bacteria-1003.
[4] See id.
[5] Fieldhouse, supra note 1.
[6] See id.
[7] Gordon, supra note 3.
[8] Alex Ouyang, Speeding up drug discovery with diffusion generative models, MIT News (Mar. 31, 2023), https://news.mit.edu/2023/speeding-drug-discovery-with-diffusion-generative-models-diffdock-0331.
[9] See id.
[10] See id.
[11] Gordon, supra note 3.
[12] See id.
[13] See id.
[14] See id.
Subscribe for News and Updates
