The eyes of the world will be on this increasingly powerful technology as the United Kingdom prepares to hold a global artificial intelligence (AI) safety summit in November 2023. However, early signs indicate the discussions will lack ambition.

In this report, the authors firstly suggest how, rather than merely focussing on harms, governments should outline a positive vision for how AI can help create public value. Secondly, they assess potential structural harms to the economy, such as subtle ways of consumer deception and runaway market dominance of a small number of players, squeezing value creating innovation by smaller firms in the real economy.

Finally, the authors propose the establishment of an Advanced AI Monitoring Hub, a technically-specialised agency that is given oversight access to what is deemed ‘systematically important AI infrastructure’.