US Stalemate Leaves AI Regulation Stalled in Washington
Despite growing concern over the risks posed by advanced AI models, neither the White House nor Congress can agree on a concrete regulatory approach. President Trump has shelved an oversight agency proposal and dismissed AI safety fears as a hoax.
Washington has spent weeks debating the need for AI regulation as concerns grow about the risks of highly capable frontier models, yet no serious action appears imminent. Over the summer, White House aides began drafting a framework for an AI oversight body after Google DeepMind cofounder Demis Hassabis proposed the idea in July, envisioning a system where leading AI labs would largely self-regulate the development of frontier models.
The Treasury Department and the White House Office of Science and Technology Policy put together a preliminary draft of such a body. But once tech executives, including former White House AI czar David Sacks and Meta CEO Mark Zuckerberg, learned of the plan, they lobbied President Trump directly against it. Since August, the proposal has remained stalled, and no new oversight plans are being seriously considered inside the administration.
Trump has recently hardened his opposition to any formal AI regulation, calling safety concerns a "hoax" in a social media post. Part of his frustration reportedly stems from a lengthy essay by Anthropic CEO Dario Amodei warning about the dangers of frontier AI, which aides felt created unnecessary political headaches.
Competing bills in Congress
The bipartisan Frontier Act, introduced by Representatives Lori Trahan and Jay Obernolte, is seen as having the broadest support among House Democrats and AI industry figures. It would let the Commerce Department place auditors inside AI labs to flag safety incidents and halt development of catastrophically risky models. The White House remains undecided, torn between accelerationists wary of regulatory overreach and national security officials concerned that pausing US development could let China pull ahead.
Separately, so-called kill-switch bills — which would let the government force a shutdown of an AI model — have bipartisan backing but limited enthusiasm from tech executives, who note such measures would do little to prevent damage from a model already acting undetected.
With no House voting days scheduled before the midterm elections and no consensus even among Democrats, the odds that any of these proposals pass in the near term remain very low.
