Washington won't regulate AI anytime soon

The White House has frozen its push for a FINRA-style oversight body after President Donald Trump cooled on the idea, and competing bills on Capitol Hill lack the votes to advance. Presidential advisers began sketching a framework for an administration-linked watchdog shortly after Demis Hassabis, co-founder of Google DeepMind, floated the concept in a July 14 essay and in briefings with senior officials. The Finance Ministry and the White House Office of Science and Technology Policy envisioned a body in which leading AI labs would self-regulate frontier-model development; they drafted an initial feasibility review.

Industry pressure worked. When a string of tech executives — including David Sacks, Trump's former AI czar, and Meta chief executive Mark Zuckerberg — heard about the move, they called Trump individually in August to register opposition. The proposal has lingered in limbo since. No new oversight proposal is under serious discussion at the White House, partly because Trump has doubled down against any visible regulation and labeled AI safety concerns a "hoax" in a Truth Social post. Trump's anger stems at least in part from a 3,822-word letter published by Dario Amodei, chief executive of Anthropic, warning of frontier-model dangers. Presidential aides saw it as a challenge that created unnecessary headaches, even though Amodei's position is considered relatively moderate within the industry.

With no White House lead, several bills have emerged in Congress but the House has no voting days left before the midterm elections, and no consensus has formed around any of them. The odds of passage are slim. The bipartisan Frontier AI Act, introduced this summer by senior Democratic Rep. Lori Trahan and Republican Energy and Commerce Committee member Jay Obernolte, commands the broadest support among Democrats and AI executives. As drafted, the bill would authorize the Commerce Department to embed auditors inside AI labs to report critical safety incidents and give the government authority to halt development if a model poses catastrophic risk.

Partisan politics is blocking everything. OpenAI has voiced public support, but the White House remains ambivalent about independent auditors; skeptics from the accelerationist camp fear regulatory capture, while security hawks worry about China. The core argument from Trump's camp: even if auditors deem a model dangerous and order it taken offline, there is no guarantee China would pause development or stop trying to reverse-engineer frontier-model weights. That would create immense pressure to lift any emergency freeze to stay ahead of China, hollowing out the whole concept of auditors pushing the Commerce Department to pull models. Yet the main obstacle to the Frontier Act remains partisan politics. Democratic leadership aides in the House say that even if the bill secured enough votes, Speaker Mike Johnson has no interest in bringing it to the floor. To sidestep the procedural hurdle, Democrats are exploring attaching AI-audit provisions to must-pass legislation such as a continuing resolution to fund the government, but that, too, remains at the exploratory stage.