Hassabis Pushes FINRA-Style US Watchdog to Vet Frontier AI

Google DeepMind CEO Demis Hassabis has spent the past week trying to convert a manifesto into an actual Washington plan. Per Bloomberg, he is lobbying the Trump administration on a proposal for a US-led standards body modelled on FINRA, the private, industry-funded regulator that polices Wall Street under SEC oversight, with a mandate to vet frontier AI models before they ship.

The shape of the pitch, first laid out in his manifesto and reported by Axios on July 14, is more concrete than the usual 'we should regulate AI' call. Frontier labs would voluntarily share models with the body up to 30 days before release, and the body would run safety testing that probes cyber, biological and deception capabilities. The board would sit majority-independent, with Turing Award winners alongside industry, government and open-source representatives, and the labs themselves would foot the bill. Hassabis told Axios he wants the thing running within 'months,' ideally before the end of 2026, and that the administration's private signals have been positive.

The reason this reads as more than another op-ed is the direction of travel inside the White House. Back in May, National Economic Council director Kevin Hassett publicly ruled out standing up any 'giant new bureaucracy to approve AIs,' a line then reinforced by chief of staff Susie Wiles and former AI and crypto czar David Sacks. Trump's June 2 executive order, framed almost entirely around cybersecurity, landed somewhere in between: it asked companies to submit new models for a voluntary 30-day government review, not the mandatory regime Hassabis wants. The Hassabis plan effectively tries to convert that voluntary window into a durable institution.

The honest caveat is that what has been reported this week is a lobbying effort and warm private signals, not a decision. Scope, statutory authority, which agency actually houses this thing, and how a private board could realistically block a US release of a foreign frontier model are all questions the reporting does not answer. Rival reactions have been friendly on the surface — Sam Altman called it 'thoughtful,' Elon Musk called it 'a good starting point for discussions' — but polite is not the same as agreeing to be gated.

If the White House does back something in this shape, the winners are the incumbent frontier labs that can afford the compliance overhead and the safety-testing ecosystem that would grow around them. The party with the strongest reason to lobby back is any smaller open-weight or foreign lab that finds 'pass the US body' quietly turning into the price of the American market.