Hassabis Proposes US Standards Body for Frontier AI
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Demis Hassabis proposed a US-based Standards Body for 'Frontier-class' AI, modeled after FINRA, in which labs would share models for review up to 30 days before release
- Hassabis described AI's potential impact as '10x of the Industrial Revolution at 10x the speed' and called the moment 'pivotal in human history' regarding AGI
- Sam Altman called the proposal 'thoughtful' and Aaron Levie said it was 'distinct from a regulatory agency,' while @bonegpt dismissed the standards body idea as 'garbage'
- Timothy B. Lee publicly challenged Hassabis's '10x Industrial Revolution' claim, arguing that 'humans in 1945 had more than 10% of all useful knowledge'
- Cross-coverage converged on a 'US-led AI watchdog' frame (The Verge, Axios, Financial Times), while Gary Marcus separately characterized the proposal as Hassabis endorsing 'preflight safety testing for AI'
Why it matters: Hassabis's proposal would create a 30-day pre-release review window for frontier AI models through an industry-led body — shifting the AI governance conversation from government regulation to voluntary lab coordination. Sam Altman endorsed it as 'thoughtful,' while critics called the standards body 'garbage,' exposing a live split over whether self-policing can substitute for binding oversight before any AGI release window opens.
