An early warning system for novel AI risks

24 mai 2023

,

New research proposes a framework for evaluating general-purpose models against novel threats