White House's 'FINRA for AI' plan stalls without Trump's buy-in
The Trump administration's most concrete attempt yet to let the AI industry police itself is stuck in limbo — a draft executive order that would create a self-regulatory body for AI companies has been circulating for weeks without winning the one endorsement that matters.
The White House has quietly circulated a draft executive order that would establish a self-regulatory organization for AI companies — and the plan has stalled. The Information's Leo Schwartz reported Thursday that administration officials floated the draft "in recent weeks," but progress has frozen while it waits on buy-in from President Trump and other key players. The order builds on a proposal from then-Google DeepMind CEO Demis Hassabis, who in July called for a public-private body to supervise AI labs modeled on FINRA, the financial industry's self-regulator. It also draws on proposals from senior officials including Treasury Secretary Scott Bessent, whose version Bloomberg reported on last month.
Under the FINRA model, AI labs would fund a private body that writes standards, audits members and disciplines violators — keeping oversight out of federal agencies while giving the industry cover against state and congressional action. That matters because the draft lands as Congress debates AI legislation and states like California push their own transparency rules. If the order dies, the administration is left with no federal mechanism at all; if it survives, it would be the first formal structure for AI self-governance in the US.
The friction is coming from inside the deregulation camp. David Sacks, the former AI czar who has left government, called the "FINRA for AI" idea "a horrible idea" on the All-In podcast last week — and his resistance is a big reason the order hasn't advanced. The irony is hard to miss: an administration that campaigned on deregulation is discovering that industry self-governance is politically trickier than pure hands-off, because it still hands power to a new institution. We covered the original pitch in August — Hassabis pitched an IAEA-style AI safety body to Trump officials — and the mechanics have shifted since: what was a private proposal is now a drafted policy document, and the open question has moved from design to whether the president will sign at all.
What to watch: whether the draft resurfaces or quietly dies — and whether the administration settles for Sacks-style hands-off over any formal structure.
Would an industry-run AI regulator actually hold labs accountable, or is self-regulation just regulation in name only? Tell us in the comments.
