Codex helps researchers search genomes for new antimicrobial candidates

OpenAI’s report profiles bioengineer César de la Fuente and his lab, which use AI to search the genomes of living and extinct organisms for molecules that could fight drug-resistant infections. The article was published September 10, 2026.

The lab’s deep-learning systems scan large biological sequence databases for patterns associated with potentially useful antimicrobial molecules. According to OpenAI, this can shrink the initial hunt from years to hours—though any promising result still requires extensive laboratory testing.

That distinction matters. Researchers must verify whether a candidate kills a target microbe, establish effective concentrations, measure effects on human cells, assess toxicity and resistance, and develop a practical manufacturing process. Regulatory review and clinical trials come later.

AI as a cross-disciplinary lab partner

ChatGPT and Codex support the work beyond candidate selection. Lab members use them to brainstorm hypotheses, write and improve code, download and prepare genome datasets, analyze results, explain unfamiliar concepts, and connect ideas across biology, chemistry, engineering, and computer science. The tools can also help researchers work in their native languages.

For AI builders, the example highlights a practical pattern: combine general-purpose assistants with specialized models and human validation. Codex can lower the programming barrier for biologists, while ChatGPT provides a shared space for organizing ideas from researchers with different expertise.

De la Fuente nevertheless stresses that AI output must be checked against experiments. The tools can accelerate discovery and broaden the search space, but they do not establish that a molecule is safe, effective, or suitable as a medicine. In life sciences, the value comes from linking faster computational exploration to reliable physical evidence.

Source: OpenAI News