A lab at the University of Pennsylvania is using AI to find new infection-fighting molecules in hours instead of years. On September 10, 2026, OpenAI shared how César de la Fuente’s team uses Codex and ChatGPT alongside its own models to search the genetic code of living and extinct organisms for candidates worth testing in the lab.
Here’s why that matters at home. Ordinary families still depend on infection medicine for an ear infection, a UTI, strep, childbirth, and surgery. When those medicines stop working as well, a routine problem gets harder to treat. OpenAI’s post notes that bacterial drug resistance was already tied to about 5 million deaths in 2021, and that toll is projected to roughly double by 2050. De la Fuente also points out that we have not had a new class of antibiotics in 50 years.
The September story sits on top of earlier lab results. On May 13, 2026, Penn Engineering described ApexGO, an AI method from de la Fuente’s group and collaborators, published in Nature Machine Intelligence. Instead of only scanning giant libraries for a lucky hit, ApexGO starts with a few imperfect molecule candidates and improves them step by step. Penn reports that most of the AI-designed molecules stopped bacterial growth in lab dishes, and 72% beat the starter versions against hard-to-treat bacteria. In mice, some of the new molecules lowered bacteria at levels similar to a last-resort hospital antibiotic. Those mouse tests were small and early. This is preclinical research, not a pharmacy drug and not a cure.
What the AI changes is speed and focus. Searching for useful molecules used to take years of trial and error. OpenAI says the lab’s approach brings the first search stage down to hours. Codex and ChatGPT help the team write code, sort data, and connect ideas across biology and computer science. Lab tests still decide what is real. De la Fuente is clear that ground-truth experiments have to check the model’s guesses.
Tools like this point toward real progress in everyday life. More candidate medicines reach the lab sooner. Doctors and families keep more options when a common infection stops responding to the usual pill. None of that replaces clinical trials or regulatory review. The direction still matters: faster early discovery, stronger leads for the science that follows, and more chances to keep ordinary infection care working for the kids and adults who need it.