Great article. I love the 'CERN for AI' idea from Gary Marcus, and also the recent frontier standards body article from Demis Hassabis feels…

Great article. I love the "CERN for AI" idea from Gary Marcus, and also the recent frontier standards body article from Demis Hassabis feels less like competing proposals than two halves of the same institution. We need an international organization with the compute and talent to pursue ambitious public interest research, but also the independence to test frontier models rapidly, develop constantly evolving benchmarks, and identify catastrophic risks before deployment (pushing to prod on a Friday, etc). The goal (lofty as it is) would be a layer of shared scientific infrastructure and accountability above it. If no country is likely to achieve a permanent monopoly on intelligence, can we build a common institution strong enough to preserve the upside, govern the danger, and ensure that the most powerful technology we have ever created becomes more than the property of whichever corporation or nation briefly leads. The US may not win the AI race against China What can we do about it? @GaryMarcus outlines next steps: https://substack.com/home/post/p-207547165

Source: Gary Marcus (X) | 2026-07-21