San Francisco-based Radical Numerics administered AI-designed bacteriophages to a human patient for the first time under compassionate use. The experimental treatment targeted antibiotic-resistant bacteria, marking a shift from laboratory testing to clinical application, though results remain undisclosed.

AI-Designed Virus Administered to Human Patient

Radical Numerics, a San Francisco-based AI lab developing bioscience models, announced this week that it has administered an AI-designed phage cocktail to a human patient under a compassionate-use arrangement

1

. The experimental treatment for bacterial infection represents the first time AI-designed bacteriophages have been used clinically on a person, shifting the technology beyond laboratory validation to actual patient care

2

.

How Omnii-1 Created Targeted Bacteriophages

Source: Mashable

Source: Mashable

The company's AI model Omnii-1 was used to design viable bacteriophages, viruses that specifically infect bacteria rather than human cells

1

. Radical Numerics modified parts of an existing virus's genetic material, particularly sections that help it recognize and attach to bacteria

3

. The AI-designed phage cocktail was tested against bacteria isolated from the patient's antibiotic-resistant infection before administration. Laboratory tests showed the cocktail eliminated the patient's antibiotic-resistant isolate in vitro

1

.

Addressing Antimicrobial Resistance Through Phage Therapy

Phage therapy isn't novel itself, but using AI to design these bacteria-targeting viruses accelerates the process significantly. Bacteriophages offer advantages over antibiotics, most notably their ability to target specific bacterial strains while sparing beneficial gut bacteria that broad-spectrum antibiotics can affect

1

. Finding the right virus for a specific infection traditionally requires searching through collections of naturally occurring viruses to identify one that can target the bacteria causing a patient's illness

3

. Antibiotic-resistant bacteria increasingly render conventional medicines ineffective, making alternatives pressing as drug-resistant infections already cause substantial mortality worldwide

1

.

Evaluating Omnii-1's Performance Against Bacterial Strains

The research tested Omnii-1's ability to predict which phages could infect Pseudomonas aeruginosa, a bacterium that can develop antibiotic resistance and cause infections in patients with burns and other wounds, as well as persistent lung infections in cystic fibrosis patients

1

. The AI model performed competitively against specialized models at predicting phage-host interactions, though there's nothing to suggest the patient was suffering from a Pseudomonas aeruginosa infection specifically

1

.

Missing Results and Transparency Concerns

Radical Numerics hasn't disclosed the patient's diagnosis, treatment details, or whether the infection was successfully cleared

3

. The company cited patient confidentiality and the need to protect their privacy as reasons for withholding information

1

2

. CEO Eric Nguyen said the team plans to share more details once they receive consent from the patient

2

. The announcement wasn't peer-reviewed or published in a scientific journal; instead, the company posted the paper on its website

2

. Observations from a single patient don't establish whether the approach is safe or effective for wider clinical application

1

.

Precedent Set by Stanford and Arc Institute Research

AI has already shown promise in speeding up phage engineering. Researchers from Stanford University and the Arc Institute created AI-designed phages using Evo 2 that targeted E. coli bacteria last year

1

. Of nearly 300 AI-designed viral genomes synthesized and tested, 16 produced functional viruses

3

. The phages included several variants that outperformed their biological predecessor in laboratory tests, though those experiments took place in laboratories with bacteria, not human patients

1

3

.

What This Means for Future Treatments

Radical Numerics hopes its work will lead to a process for designing and building phages targeted to individual bacterial infections

1

. Harvard microbiologist Michael Baym told The New York Times his reaction was cautiously optimistic: "If it works, that's incredible," though he added, "My impulse is to be skeptical"

3

. The first human administration marks an important milestone in translating AI-designed biology from computational predictions and laboratory validation to clinical administration

1

. While phage therapy can last anywhere from a couple of days to a few weeks, doctors also have to monitor patients for months afterward to ensure the infection was actually cured and won't return

2

.

© 2026 TheOutpost.AI All rights reserved