Experts warn open data may help AI design dangerous pathogens
Leading researchers argue that while open biological data drives scientific progress, unrestricted access to specific pathogen datasets creates severe biosecurity risks in the age of advanced AI. These models can predict mutations and design enhanced pathogens, potentially accelerating the creation of biological threats. The core conclusion is that a balance must be struck: maintaining openness for general research while implementing rigorous safeguards for high-risk data to prevent misuse by malicious actors. To address this gap, the authors propose a new governance framework that categorizes biological data into tiers based on biosecurity risk. This tiered system allows everyday biological information to remain freely accessible while applying strict controls to data that could enable the design of dangerous viruses. By defining which datasets pose meaningful risks before they are widely available to AI developers, this approach aims to preempt the accidental or intentional development of bioweapons through open-source AI tools. This article is highly relevant to open data communities as it challenges the assumption that all scientific data should be universally and unconditionally open. It highlights the critical need for nuanced data governance, introducing concepts like watermarking and behavioral biometrics to verify legitimate users. The proposal urges the community to adopt consistent, expert-backed rules for sensitive information, demonstrating that responsible open data practices must evolve to include robust security measures alongside transparency.
Source: euronews.comPublished on 2026-02-18