Washington Won’t Be Overseeing AI Regulations in the Near Future

For weeks, Washington has been buzzing with discussions about the need for AI regulation as worries grow over the potential risks of powerful frontier AI models going off track.

However, there’s little reason to believe any single proposal will garner support from the White House, where officials have halted efforts to create a FINRA-like regulatory agency after President Donald Trump expressed his disapproval. Likewise, on Capitol Hill, lawmakers have conflicting proposals that lack the necessary votes for passage.

White House aides began outlining a framework for an AI oversight agency tied to the administration shortly after Google DeepMind’s cofounder Demis Hassabis suggested the idea in a July 14 essay and while briefing senior officials, according to insiders.

The Treasury Department, along with the White House’s Office of Science and Technology Policy, envisioned an entity where leading AI laboratories would self-regulate the advancement of frontier models, and they drafted an initial proposal to explore its implementation.

However, when a host of tech leaders—including Trump’s former AI advisor David Sacks and Meta CEO Mark Zuckerberg—became aware of the initiative, they each contacted Trump in August to voice their opposition. The proposal has since been stalled.

Since that time, there have been no serious new AI oversight proposals considered in the White House, as sources indicate, particularly since Trump has recently reiterated his rejection of any form of regulation, labeling AI safety concerns a “hoax” in a recent Truth Social post.

Trump’s discontent has, at least partially, been fueled by a lengthy 3,822-word commentary by Anthropic CEO Dario Amodei regarding the dangers of frontier AI. According to one source, Trump’s aides regarded it as creating unnecessary anxieties that complicated matters for the administration, despite it reflecting a relatively moderate stance within AI discussions.

In the absence of leadership from the White House, several bills have emerged in Congress. However, with no voting days remaining on the House calendar until after the midterm elections and a lack of consensus around any single bill, the chances of passing anything are extremely low.

Independent Auditors

The bipartisan Frontier Act, introduced this summer by House Democratic senior whip Lori Trahan and Republican House Energy and Commerce Committee member Jay Obernolte, is viewed as having the most backing within the House Democratic caucus and among AI executives.

If passed, the bill would allow the Commerce Department to embed auditors at AI laboratories, enabling them to report significant safety incidents and allowing the administration to halt development if a model is determined to pose a catastrophic risk.

Despite public endorsement from OpenAI, the White House has remained uncertain about endorsing the idea of independent auditors, facing skepticism from AI accelerationists concerned about regulatory capture and from national security advocates wary of China.

Critics from Trump’s circle argue that even if auditors conclude that a model is dangerous and should be paused, there’s no assurance that China would also halt its development or refrain from attempting to reverse engineer frontier AI models.

This would create intense pressure to lift any emergency halts in development to maintain a competitive edge over China, undermining the entire rationale for having auditors to compel the Commerce Department to take models offline.

Nevertheless, the primary challenge for the Frontier Act continues to be partisan politics. House Democratic leadership aides inform Inner Loop that even if the bill secures enough votes for passage, Republican Speaker Mike Johnson shows no interest in bringing it forward for a vote.