US startup Radical Numerics announced Omnii Defense, a model designed to flag possible biological threats without an internet connection. Access is restricted to vetted external testers while laboratories and other evaluators assess it. The system produces predictions from genetic sequences. An independent researcher cautions that finding a signal remains distinct from determining its actual consequences or engineered origin.
Radical Numerics, a US startup developing biological AI, announced Omnii Defense on November 11. The model is intended to flag potential biological threats without an internet connection.[1], [2]
The company describes portable use in field settings with poor connectivity. Access is limited to vetted external testers, while national laboratories, government agencies and other independent evaluators assess the system.[1]
The system produces predictions about genetic sequences, including possible disease-causing potential, possible AI origin and biological classification. Radical Numerics describes its approach as learning relationships between sequence patterns and biological function. The company contrasts this with tools that look for matches in libraries of known organisms. Its broader Omnii model family also covers several forms of biological information, rather than DNA alone.[1]
Co-founder Eric Nguyen worked on earlier biological AI models during doctoral research at Stanford University before helping establish Radical Numerics. Wired also spoke to Bronwyn MacInnis, a researcher at the Broad Institute, a biomedical research institution. She said technology can contribute to vigilance, while warning that detecting a signal is a substantial step away from determining actual consequences or whether a threat was engineered. Omnii Defense remains at the external-testing stage.[1]
