Scientists have reported the first use of viruses designed by artificial intelligence (AI) to treat a human patient, marking a potential advancement in the fight against drug-resistant bacterial infections. The treatment involved administering an AI-created bacteriophage, or phage, to combat a severe infection caused by Escherichia coli bacteria resistant to existing antibiotics and phage therapies.
The startup company Radical Numerics, which developed the AI-designed viruses, announced last week that it had used its technology in a clinical trial authorized by the U.S. Food and Drug Administration (FDA). However, the company declined to disclose specific details about the patient or the medical team involved, citing confidentiality agreements. Radical Numerics posted a technical report outlining the effort on its website, but the document has not undergone peer review or been published in a scientific journal.
Phage therapy—using viruses that infect and kill bacteria—has existed for over a century but has gained renewed interest in recent years amid the rise of antibiotic-resistant bacteria globally. Traditionally, phage therapy requires isolating naturally occurring phages that can target a patient’s bacterial strain, a process that is often time-consuming and labor-intensive.
To overcome this challenge, researchers have explored AI models capable of designing new phages tailored to specific bacterial infections. In 2025, scientists at the Arc Institute in Palo Alto, California, created an AI system called Evo 2, which generated hundreds of thousands of synthetic virus genomes by analyzing millions of DNA sequences. Of these, several viable phages were successfully synthesized in the lab.
Several members of that research team, including Radical Numerics CEO Eric Nguyen, later launched the company and developed a new AI model named Omni. Unlike its predecessor, Omni integrates diverse biological data such as protein structures and gene regulation mechanisms to design viruses. Using Omni, the company engineered variations of the T7 phage, which infects E. coli by attaching to the bacterial surface with specialized leg-like structures.
The AI-generated modifications altered the shape of T7’s legs with the goal of enabling the phages to bind to drug-resistant E. coli strains. After confirming the engineered phages’ ability to infect bacteria in vitro, Radical Numerics collaborated with physicians treating a patient whose infection had not responded to conventional treatments.
External experts have expressed cautious interest in the development. Michael Baym, a microbiologist at Harvard University, noted the potential significance if the treatment proves effective but emphasized the need for detailed scientific evaluation once clinical data are available. Brandon Berryhill, an evolutionary microbiologist at Uppsala University, pointed out that success with AI-designed phages does not necessarily clarify why phage therapies work or fail.
Dr. Nguyen said the company aims to reduce the time needed to develop tailored phage therapies from weeks or months to just days, potentially accelerating treatment options for infections resistant to all other interventions. The outcome of the current trial remains pending, and medical professionals typically monitor phage therapy patients over extended periods to detect possible relapses. If successful, this approach could represent a novel application of AI in precision medicine targeting bacterial diseases.