An early warning system for novel AI risks

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

Source: Read More