A Startup Wants to Fight AI Bioweapons With AI
A startup has unveiled a new artificial intelligence model designed to flag natural and AI-generated pathogens, aiming to bolster biodefense.
Radical Numerics, a San Francisco-based company founded in 2025, has developed Omnii Defense, an edge model designed to identify emerging biological threats.
The tool runs on a laptop and a USB gene sequencer, functioning without an internet connection to support researchers in the field.
It assigns a genome a "threat score" between zero and one, combining the likelihood of unnaturalness with functional predictions to gauge risk.
AI Model's Capabilities
Omnii Defense makes multiple predictions about a genetic sequence, including the odds of it being pathogenic or AI-generated and its taxonomic group.
Unlike earlier tools like the federally funded Felix, which rely on cataloged matches, AI models can learn patterns in biological function.
This allows them to recognize similarities even when genetic sequences have changed, identifying potential risks from novel or modified pathogens.
Omnii is a "step change."
Testing and Field Use
The company tested Omnii by removing the Bundibugyo virus, which is driving an Ebola outbreak in East Africa, from its training data.
Even with the virus removed, the model flagged 33 percent of the sequences as potentially pathogenic, demonstrating its ability to detect novel threats.
This capability is crucial for public health officials to respond to emerging biological threats before they can spread widely.
Company Controversy
Radical Numerics recently faced scrutiny after omitting a partnership with Tabula Bioscience in a public announcement about a custom bacteriophage treatment.
The omission led Harvard geneticist George Church, an adviser to both companies, to resign from his role at Radical Numerics last week.
Related
