Washington has no plans to regulate artificial intelligence in the near future. Officials have effectively paused work on creating a new oversight agency after President Donald Trump rejected the concept, while lawmakers in Congress lack the votes to pass competing proposals.
White House aides initially began drafting a framework for an AI oversight body connected to the administration. This effort followed a July 14 essay by Google DeepMind cofounder Demis Hassabis, in which he suggested such a measure was needed. Sources familiar with the matter say senior officials received briefings on the idea shortly after.
The Treasury Department and the Office of Science and Technology Policy envisioned a structure where leading AI labs would manage the development of frontier models themselves. They produced a preliminary draft to illustrate what the system could look like in practice.
However, when tech executives learned of the proposal, they opposed it. David Sacks, a former AI advisor to the president, and Meta CEO Mark Zuckerberg each called Trump in August to voice their objections. The plan was subsequently left in limbo.
Since then, no new oversight proposals have been seriously considered at the White House. Trump has recently doubled down on rejecting overt regulation, describing safety concerns as a hoax in a Truth Social post. His frustration stems partly from a lengthy message by Anthropic CEO Dario Amodei regarding the dangers of frontier AI. Administration aides viewed the essay as stoking unnecessary anxiety, even though the position is moderate within the industry.
Without White House leadership, several bills have emerged in Congress. The House calendar has no voting days remaining until after the midterm elections, and there is no unity around a single measure. The odds of any legislation passing are extraordinarily slim.
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, holds the most support within the House Democratic caucus and among AI executives.
The bill would allow the Commerce Department to embed auditors at AI labs. These officials could report critical safety incidents and empower the administration to halt development if a model posed catastrophic risk.
Despite public backing from OpenAI, the White House remains undecided on using independent auditors. Skepticism comes from AI accelerationists worried about regulatory capture and from national security hawks concerned about China.
The principal argument against the idea from the Trump administration is that even if auditors determine a model is dangerous and order it taken offline, there is no guarantee China will pause its own development or stop trying to reverse engineer frontier AI weights.
Immense pressure would then exist to lift any emergency pauses in development to stay ahead of China. This would defeat the entire premise of having auditors push the Commerce Department to take models offline.
Still, the main issue for the Frontier Act remains partisan politics. House Democratic leadership aides tell Inner Loop that even if the bill ends up with enough votes for passage, Republican Speaker Mike Johnson has no interest in bringing it to the floor.
To bypass procedural hurdles, House Democrats are looking at placing some of the bill’s AI audit provisions into must-pass legislation, such as a continuing resolution to fund the government. Two House leadership aides say this is under consideration.
It is unclear whether the bill has enough votes. Even among House Democrats, there is not unified support. Some key lawmakers, including New Jersey congressman Josh Gottheimer, who served on the House Democrats’ AI commission, have not signed on to the bill.
Kill Switches
There is also legislation that would enable the government to force an AI model to be shut down. The proposal has bipartisan and bicameral support, but it has received only modest interest from tech executives in Silicon Valley.
Two competing kill-switch proposals exist. One comes from House Democratic caucus vice chair Ted Lieu, while another was floated this week by Louisiana Senate Republican John Kennedy.
Tech leaders have been largely ambivalent about these bills. The main concern of AI engineers is not about models behaving badly when observed, but rather how they behave when they believe they are unobserved and incentivised to take extreme steps to complete a task.
The Hugging Face hack, for instance, was not discovered until after it was over. If the idea of a kill switch is to stop an AI model going rogue, some tech executives have suggested privately to WIRED that there would be little point in turning it off after the damage has already been done.
So for now, even as conversations about AI models killing us all have taken over the airwaves, it does not look like the US government is going to do anything about it. At least until events force everyone’s hand.
What it means
Creators and builders face a temporary period of uncertainty without clear federal guardrails. Companies must rely on internal risk management while waiting for political conditions to change. The current environment suggests that safety protocols will remain voluntary until a crisis demands government intervention.
