Ex-FDA Scientist Says Biotech Guides AI Oversight
A former FDA regulatory scientist argues that the US should regulate AI the way it regulated biotechnology: by using existing federal authority, assigning oversight by risk and use case, and avoiding a fresh layer of new legislation. The post presents biotech’s Coordinated Framework as a proof point that flexible, centralized governance can support innovation while still managing real harms, and points to two open-access working papers proposing NIST-led frontier model oversight, domain-agency regulation for applications, and a pre-deployment review process modeled on GRAS notifications.
Hot take: this is one of the more credible “regulate AI like X” arguments because it comes from someone who actually helped build a modern regulatory framework, not a generic policy pundit.
- –The strongest idea here is jurisdictional layering: frontier models get one oversight lane, downstream applications get another, and sector regulators keep their existing authority.
- –The weakest point is execution risk: AI diffuses faster than biotech, so a framework that works on paper can still become under-enforced or captured in practice.
- –The GRAS-style pre-deployment review is the most interesting part because it suggests a scalable middle ground between self-certification and full premarket approval.
- –Even if you disagree with the policy prescription, the post is useful because it reframes AI governance as an institutional design problem rather than a purely ideological one.
DISCOVERED
45d ago
2026-04-24
PUBLISHED
45d ago
2026-04-24
RELEVANCE
AUTHOR
MeatHumanEric