Researcher uses Codex and ChatGPT to hunt new antimicrobial molecules

César de la Fuente's laboratory is applying OpenAI's Codex and ChatGPT to search living and extinct genomes for novel antimicrobial candidates — molecules that could fight drug-resistant infections, one of medicine's most pressing threats. OpenAI documented the work as a case study in applied AI for biomedical discovery.

Mechanically, the approach uses LLMs to mine vast genomic datasets, including sequences from extinct organisms, for peptide and molecule candidates with antimicrobial properties. Codex handles the computational pipeline while ChatGPT assists reasoning over biological data, compressing a search space that would take human researchers years to traverse manually.

The application stands in pointed contrast to this week's darker AI-bio news: Anthropic disclosed blocking users attempting to use Claude for biological weapons development. De la Fuente's work illustrates the dual-use tension at the heart of AI-in-biology — the same capabilities that could accelerate antibiotic discovery could, misused, aid pathogen design. OpenAI's decision to spotlight the constructive application is partly a narrative counterweight.

Competitively, AI-for-science is becoming a prestige battleground. Google DeepMind unveiled AlphaGenome Atlas this week (scoring every possible DNA letter change), and OpenAI recently claimed a Navier-Stokes math breakthrough. Demonstrating real biomedical impact helps labs justify their compute spend and safety postures. The caveat: candidate molecules are a long way from validated drugs — watch for wet-lab confirmation and whether any de la Fuente candidates advance toward clinical testing.