Daniel Reuter

I'm an ML and security researcher working on protocols by which datacenters can demonstrate their compliance with future international agreements on AI development.

I'm currently a research affiliate at the Oxford Martin AI Governance Initiative and a MATS scholar.

Previously, I did empirical work in Raj Chetty's lab, at the Fed, and with Joseph Stiglitz, and studied math and economics in college.

Some recent work:

[ICML TAIGR 2026] Inference verification despite nondeterminism
[ICML TAIGR 2026] Defending against model exfiltration using inference verification

You can reach me at dreuter14@gmail.com.