Radical Numerics, a lab based in San Francisco, has revealed that it administered a model called Omnii-1, which designs bacteriophages, to a patient suffering from a drug-resistant infection. This marks a significant step, as it’s reportedly the first time an AI-designed organism has been used in a human medical context. However, the company hasn’t disclosed the results of this treatment to the public yet, leaving many questions unanswered.
The bacteriophages were developed by modifying the gene responsible for T7’s tail fiber—a protein essential for the virus to attach to bacteria. But, notably, no one outside of Radical Numerics knows if the treatment was effective since the results haven’t been peer-reviewed or made public.
Radical Numerics has stated that its AI-designed viruses were given to a patient, but details about what ensued afterward remain sparse. The company announced this in October 2026, explaining that this treatment was part of a physician-led compassionate-use pathway approved by the FDA. If verified, this would be a groundbreaking achievement in pharmaceuticals.
AI’s Role in Phage Therapy
Bacteriophages, or phages, are viruses that specifically target and infect bacteria, leaving human cells intact. They’ve been recognized as a form of therapy for over a century and have seen renewed interest as antibiotic resistance grows. According to the World Health Organization, phage therapy is generally safe since it can target a specific bacterial strain while preserving the rest of the microbiome.
However, the challenge is that each phage is designed to recognize a very specific set of bacterial receptors, making the treatment process complex. Clinicians typically go through existing phage libraries and environmental samples, often resorting to engineering new phages when necessary—especially crucial in urgent cases like sepsis.
The situation surrounding antibiotic resistance is dire. The Global Burden of Disease project has found that antimicrobial resistance was linked to approximately 4.7 million deaths in 2021, with predictions showing that nearly 40 million deaths could occur between 2025 and 2050 due to this issue.
Founded by the creators behind the Evo genome models, Radical Numerics aims to improve phage design. Their earlier work yielded 700,000 candidate genomes, resulting in a limited number of viable phages. Omnii-1 represents the next evolution; instead of creating complete genomes, this approach focuses on refining the targeting mechanism. The team’s work involved modifying a well-known E. coli phage and rewriting a specific protein related to receptor-binding.
Unpublished Results and Future Implications
While the company did test its phage cocktail against bacteria from a real patient’s infection—and it performed well in vitro—the specifics of the patient’s outcome remain unknown. The patient had an E. coli infection resistant to both antibiotics and all known phages, which is undoubtedly a critical scenario in infectious disease management.
Since the announcement, Radical Numerics has held back on details regarding the clinical outcome, citing patient privacy concerns. Additionally, although they shared a technical report on their website, it hasn’t undergone peer-review yet. The treatment is still proprietary, and the company hopes to eventually commercialize its findings.
As of October 29, after the announcement, reports for infectious disease specialists indicated no updates regarding the clinical outcome or any peer-reviewed research. Their technical write-up remains the only available source of information. In fairness, the company has acknowledged that single-patient cases can’t establish the safety or effectiveness of a treatment—something that sets a cautious tone amid the excitement.
Experts have expressed a mix of intrigue and skepticism regarding the treatment’s outcomes. For example, microbiologist Michael Baym remarked that while the results could be extraordinary if successful, he is hesitant to get too optimistic until more thorough evaluations can be made.
The commercial aspect of this endeavor is more straightforward; the process of finding phages can be lengthy, and Radical Numerics believes AI can significantly expedite this design phase. Ottawa Hospital infection specialist Marisa Azad, who collaborates with the company, mentioned that traditional phage hunting is indeed time-consuming, underscoring the potential for AI to revolutionize treatment timelines.
Nevertheless, there are valid concerns about the implications of such technology. The ability to create tailored viruses raises ethical questions about biosafety and containment measures needed to mitigate risks. Radical Numerics is aware of these issues and has committed to implementing safety protocols to handle this powerful technology responsibly.
The concept of using engineered phages isn’t entirely new; there was a documented case in 2019 where modified phages were successfully used against a resistant infection. However, the uniqueness of Radical Numerics lies in the AI-driven design process and the corresponding ethical considerations it invokes rather than the therapeutic idea itself.
As AI continues to advance, it’s evolving from merely analyzing biological processes to actually creating them—ultimately leading to clinical applications. This is a substantial leap from scenarios where AI just assists in proposing treatment hypotheses or diagnostics. For now, though, we are left in suspense regarding this groundbreaking experiment, unsure of how it unfolded or when more information might emerge. Should it prove successful, however, the ramifications for medicine could be significant and far-reaching.
