Joshua Tan, an American researcher at the University of Oxford, makes the case for a world of multiple safe AGIs, overseen by an institution that is, in effect, the practical realisation of Yoshua Bengio’s recent proposal: a multilateral network of publicly-funded AI laboratories that prioritize humanity's safety and well-being. Tan then focuses on how to establish such a network, as well as on resolving some operational and incentive problems he sees with Bengio's original proposal.