Radical Numerics unveils Omnii Defense to flag AI-made pathogens

Radical Numerics was founded in 2025 by Eric Nguyen and cofounders who worked on the open-source DNA foundation models Evo and Evo-2 during doctoral research at Stanford University. The company recently rolled out Omnii, a genome language model built on biological information beyond DNA, and in October announced it had used Omnii to create custom bacteriophages targeting a drug-resistant strain of E. coli, a cocktail used to treat a sick patient with special permission from the Food and Drug Administration.

That October announcement did not disclose the involvement of another AI biotech startup, Tabula Bioscience, and the omission led Harvard geneticist George Church, an adviser to both companies, to resign from his role at Radical last week. Omnii Defense is the company's latest tool, and its approach differs from earlier biodefense tools such as the federally funded Felix, which compares genetic sequences against libraries of known organisms. In one experiment, the company removed Bundibugyo virus from the training data, first at the species level and then its whole genus, and the model still flagged 33 percent and 22 percent of viral sequences as pathogenic, respectively, while non-AI tools gave essentially no indication of potential risk.

Bronwyn MacInnis of the Broad Institute of MIT and Harvard said vigilance about what AI enables is warranted, but cautioned that detecting a signal is still a big leap from determining whether a pathogen has real consequence or whether a threat was engineered.