Reader

An early warning system for novel AI risks

2023-05-25 00:00:00 +0000 UTC | Google DeepMind Blog | Default

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