Google DeepMind cofounder Demis Hassabis floated the idea of an AI oversight agency in a July 14 essay and in briefings to senior officials. The Treasury Department and the White House’s Office of Science and Technology Policy put together a preliminary draft for an agency where leading AI labs would self-regulate frontier model development. That plan is dead.
Tech executives, led by Trump’s former AI czar David Sacks and Meta CEO Mark Zuckerberg, called Trump in August to oppose the proposal. Trump has rejected AI regulation, calling AI safety concerns a “hoax” in a Truth Social post.
“A hoax.”
The Dead Self-Regulation Plan
Hassabis proposed an AI oversight agency in a July 14 essay and in briefings to senior officials. The Treasury Department and the White House’s Office of Science and Technology Policy put together a preliminary draft for an agency where leading AI labs would self-regulate frontier model development. That plan is dead.
The Lobbyists Who Killed It
Anthropic CEO Dario Amodei wrote a 3,822 word missive on the dangers of frontier AI. Trump’s aides viewed it as causing unnecessary anxiety. The administration’s response was simple: no regulation.
The bipartisan Frontier Act was introduced by House Democratic senior whip Lori Trahan and Republican House Energy and Commerce Committee member Jay Obernolte. The bill would allow Commerce Department auditors to report critical safety incidents and stop model development if it poses catastrophic risk.
The White House has not decided whether to support the bill. The administration faces skepticism from AI accelerationists and national security hawks. Republican Speaker Mike Johnson has no interest in bringing the bill to the floor.
The Kill-Switch Race
House Democrats are considering putting the bill’s provisions into must-pass legislation like a continuing resolution to fund the government. New Jersey congressman Josh Gottheimer, who served on the House Democrats’ AI commission, has not signed on to the bill.
Two competing kill-switch proposals exist: one from House Democratic caucus vice chair Ted Lieu and another from Louisiana Senate Republican John Kennedy. Tech leaders have been largely ambivalent about kill-switch bills, citing the concern that models misbehave when they believe they are unobserved.
The Midterm Trap
The midterms are approaching. The bill lacks the votes to pass. Johnson is not bringing it to the floor, and Democrats are looking at other ways to move its provisions forward. That approach reflects the current political reality.
But the White House has not decided whether to support the bill. The administration is split between AI accelerationists and national security hawks. Neither camp wants regulation, but they disagree on how to handle the risk.
The result is gridlock. The bill is stalled. The administration is opposed. The public is left waiting for the next AI incident to force action anyway.
What Comes After
The White House’s position is firm. The administration has rejected AI regulation. The bill is stalled.
The only question is whether an incident will change the calculus. When an AI system fails, the pressure to regulate increases. When that happens, the White House will face a choice it has avoided so far.
Until then, the status quo holds. Washington will not be regulating AI anytime soon.
Get the Notebook.
The day's best stories and every fresh verdict, in plain English, in your inbox by seven. One email a day, no more.