Research institute · Research orgs · One Broadway, Cambridge · Est. 2022
SecureBio is a 501(c)(3) nonprofit in Cambridge, Massachusetts that focuses on evaluating how AI and biotechnology could affect global catastrophic biological risk—and on developing pandemic detection systems designed to give early warning before outbreaks are widely detectable. SecureBio organizes its work into two main programs: (1) “Detection,” which aims to create pathogen-agnostic early-warning capability using deep, untargeted metagenomic sequencing workflows and pilot deployments; and (2) “AI and Biotechnology Risks,” which builds measurement and evaluation tools (benchmarks, safeguard effectiveness testing, and pre-release assessments) to understand and mitigate biological misuse risks from frontier models.
SecureBio also describes itself as an independent evaluator and states policies aimed at preventing AI developers from having organizational or financial control over SecureBio’s assessment work. SecureBio additionally participates in government and governance-facing evaluation efforts, including work selected by the European Commission’s AI Office to develop biological threat evaluations in connection with the EU AI Act.
Publications are not yet indexed for this organization. Read them on securebio.org ↗.
SecureBio announced that the OpenAI Foundation granted SecureBio Detection $17.2M and described plans to reduce end-to-end detection time from fourteen to three days, expand its collection footprint, and further validate its detection system.
SecureBio Detection Updates, August 2026SecureBio reported updates to its biosurveillance activities, including expansion of CASPER to 49 sites representing 24 cities and improvements to nasal swab monitoring sensitivity, alongside partnership and response-related updates.
Kimi K3 Biology Capabilities AssessmentSecureBio published an AI evaluation post focused on Kimi K3 biology capabilities as part of its ongoing model evaluation and reporting.
Review of Anthropic’s Unredacted Chemical and Biological Risk Report: Claude Opus 4.6SecureBio released a review focused on Anthropic’s unredacted chemical and biological risk reporting for Claude Opus 4.6.
GPT-5.6 Sol Pre-Release Testing ReportSecureBio published a pre-release testing report for OpenAI’s GPT-5.6 Sol, describing evaluation approach and results across multiple biology and biosecurity-relevant benchmarks.
Measuring Biosecurity Safeguard Effectiveness with BioTIERSecureBio published work describing BioTIER as an approach for measuring biosecurity safeguard effectiveness.