Google DeepMind CEO Demis Hassabis endorsed a coordinated slowdown in frontier AI development, citing risks from superintelligent AI systems that currently lack control mechanisms. However, Peter Wildeford argues that Hassabis's proposal, which focuses on commercial release and suggests a FINRA-like self-regulatory model, is insufficient as major risks occur pre-commercialization and such a model is prone to industry capture without a strong government regulator like the SEC.
The debate over AI regulation intensifies as frontier labs acknowledge existential risks, but proposed self-governance models face skepticism over effectiveness and industry capture.