A Startup Wants to Fight AI Bioweapons With AI (www.wired.com)

Radical Numerics, a startup founded by researchers behind open-source DNA models, has launched Omnii Defense, a new AI-based tool aimed at detecting both natural and AI-engineered pathogens. Unlike traditional biodefense systems that rely solely on known genetic sequences, Omnii Defense utilizes machine learning to make predictions based on patterns in biological sequences, allowing it to assign a "threat score" to genomes based on their likelihood of being pathogenic or artificially created. This innovation is particularly notable as it can operate offline, making it accessible for field research in areas with unreliable internet connections. The significance of Omnii Defense arises from growing concerns among experts about the potential misuse of AI in biological warfare. While many scientists deem AI-driven bioweapons a low-likelihood threat, Radical Numerics aims to bolster biothreat detection amidst fears of an arms race in biodefense. Early experiments demonstrate that Omnii can identify viral sequences with a surprising accuracy, even when the original pathogen data is removed from its training set, showcasing its potential to outperform existing tools like Finding Engineering-Linked Indicators. This cutting-edge advancement not only positions Radical Numerics at the forefront of biodefense technology but also opens discussions about the balance between harnessing AI for security while remaining vigilant against its potential misuse.