Hassabis Calls for US-Led AI Standards Body Like FINRA
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Demis Hassabis, CEO of Google DeepMind, proposed a US-based Standards Body for "Frontier-class" AI modeled after financial regulator FINRA, with labs required to share models for review up to 30 days before release
- Hassabis framed the proposal as responding to "a pivotal moment in human history" tied to the development of AGI, publishing the full framework on Substack as "A Framework for Frontier AI and the Dawning of a New Age"
- The proposal drew near-simultaneous coverage from The Verge, Axios, TechCrunch, the Financial Times, The Information, Business Insider, Fortune, and the Associated Press-linked Decoder, with most outlets framing it as a call for a "US-led" or "global AI watchdog"
- Gary Marcus, a long-time DeepMind critic, broke from his usual posture to endorse the preflight safety testing component, writing "Demis Hassabis endorses preflight safety testing for AI"
- Tech leaders across the industry engaged on X, including Satya Nadella (Microsoft), Sam Altman (OpenAI), Chamath Palihapitiya, and Jack Dorsey, signaling rare cross-faction attention to a regulatory proposal
- Hassabis's proposal lands as frontier AI labs face no formal pre-release review requirement in the US, making the 30-day window a concrete new mechanism rather than a vague call for oversight
Why it matters: A FINRA-style body would impose a concrete 30-day pre-release review checkpoint on US-based frontier AI labs — a binding constraint that doesn't currently exist. The proposal's rare convergence of endorsement from Google DeepMind's CEO and longtime critic Gary Marcus, plus engagement from OpenAI's Sam Altman and Microsoft's Satya Nadella, suggests the pre-release review mechanism has unusually broad cross-faction traction compared with prior AI regulation proposals.
