My project examines the rapidly evolving intersection of artificial intelligence, synthetic biology, and national security, with a focus on the biosecurity risks posed by emerging generative biology tools. Recent advances in protein language models and AI-enabled protein design have dramatically accelerated our ability to generate novel biomolecules, offering transformative opportunities for drug discovery, vaccine development, and global health. However, these same capabilities introduce the possibility of creating harmful synthetic pathogens.
To address this dual-use challenge, my work evaluates the limitations of current biosafety screening systems, particularly inference-stage computational filters used to detect dangerous biological sequences. Through technical analyses of cutting-edge predictive models—including AlphaFold 3, AF3Complex, and SpatialPPIv2—I have shown that these tools frequently fail to identify known viral-host interactions, even for well-studied pathogens like Hepatitis B and SARS-CoV-2.
Building on these findings, I develop policy and governance frameworks that prioritize resilient, response-oriented infrastructure, rapid experimental validation pipelines, and careful distribution controls for high-risk generative models. My work aims to ensure that AI-driven biological innovation can advance safely, preserving the enormous potential benefits for human health while mitigating risks to global security.