About
My research focuses on assessing whether AI systems might meaningfully lower the barriers to weaponizing biology. My motivation is to reduce the risk of it happening.
I began this as a Pivotal Research Fellow, with mentorship from SecureBio. My first project was on automating adversarial elicitation — or, seen from another angle, measuring refusal robustness within Inspect. When a model won’t answer a hazardous question, how do you get around it? I suspect chemical and biological risk needs to adopt the concept of attacker intuition from our colleagues in cyber.
Before this I was an mRNA scientist, doing wet-lab R&D. I hold an MSc in Medical Science from Boston University (2023) and a BSc in Biochemistry from UMass Amherst (2020).
The best way to understand how I think is to read my writing — here, or on my Substack.
Find me
- Professional: LinkedIn
- Academic: Google Scholar · ORCID 0009-0007-8578-2687
- Code: GitHub: gvmfhy
- Email: via LinkedIn