Study of 52K Preprints Reveals Widespread Dual-Use Biosecurity Risks

A new systematic analysis by Vasudha Sharma and colleagues (arXiv:2605.28843) presents the first large-scale empirical study of dual-use research of concern (DURC) on open preprint servers. Using a hybrid pipeline of lexical filtering and large language model evaluation, the team screened approximately 52,000 bioRxiv preprints from 2024-2025, scoring metadata across nine DURC, three PEPP (pandemic pathogen potential), and five governance categories aligned with U.S. and Australia Group oversight frameworks.

Their analysis reveals that dual-use-adjacent knowledge — information that could be misused to create biological threats — is routinely present in openly accessible titles and abstracts, often exceeding established risk thresholds even in studies with legitimate public health objectives. The researchers emphasize that while this captures surface-level information diffusion, it does not measure actual operational capability or downstream misuse. They argue that institutional review processes, funding requirements, and preprint platform policies must evolve to incorporate proactive, metadata-level monitoring. Ultimately, they propose a pragmatic framework that harmonizes controlled-access mechanisms for high-risk methodologies with open summaries of scientific contributions, aiming to govern AI-accelerated biology at scale without compromising scientific transparency.