The science and practice of proportionality in AI risk evaluations.

Mougan, Carlos; Morlock, Lauritz; Aguirre, Jair; Black, James R M; Brauner, Jan; Campos, Simeon; Dev, Sunishchal; Llorca, David Fernández et al. · Science · 2026

expert_opinion · Level V

Where this comes from

Abstract

AI evaluations should provide meaningful risk information without imposing excessive burden.