President Trump signed an executive order giving the government optional early access to the most powerful AI models, then floated the far bigger idea of the government taking financial stakes in the AI labs.
AI has officially become front and center in Washington, and the policy ground is shifting fast.
President Donald Trump signed an executive order titled "Promoting Advanced Artificial Intelligence Innovation and Security that lets frontier AI labs opt into a federal process to have their most powerful models designated as "covered frontier models." If a model qualifies, the lab gives the government 30 days of early access before public release, so agencies can check for cybersecurity risks. The White House stressed this is not oversight of every model, only ones that represent a real jump in cyber capabilities, and the order says it creates no licensing or pre-approval requirement.
The bigger headline came two days later: Trump said he would likely meet with AI companies to discuss a federal "partnership" that would let the American people profit from their success, according to Politico. The idea of the government taking an equity stake or ownership share in AI labs has been floated by OpenAI, discussed by Anthropic and xAI, and pushed hardest by U.S. Sen. Bernie Sanders, who in a New York Times op-ed proposed a one-time 50% tax on the largest AI companies, paid not in cash but in stock. The talks are early, the legal mechanism is unclear, and it might not happen at all.
SmarterX founder and CEO Paul Roetzer discussed the order and the ownership debate on Episode 218 of The Artificial Intelligence Show.
30 days - Early-access window for the government to review AI models, reduced from 90 days
50% - Government ownership stake proposed in Sanders' bill
269 pages - Length of a separate bipartisan House AI bill
3 years - How long proposed House legislation would preempt state AI laws
$100 million - Annual funding (for three years) the House bill proposes for a new AI standards and innovation center
Optional in name only? The order calls the early-access program voluntary, but Roetzer is skeptical of what that word means coming from the government. "The way the government does voluntary is like, 'Yeah, you can choose not to participate, and then we can choose not to give you government contracts,'" says Roetzer. "So it's like wink, wink voluntary." The administration also went out of its way in an accompanying fact sheet to insist it is the most innovation-friendly in history, a balancing act driven by the need to look strong in the race for AI leadership against China without appearing to stifle the industry.
Bills are about to pop up everywhere. Two days after the order, a bipartisan House group unveiled a 269-page framework, summarized by Axios, that would:
preempt state AI laws for three years
create a new Center for AI Standards and Innovation
protect AI whistleblowers
require large developers to file risk plans before releasing new models
Roetzer's read is that this is just the beginning. "Bills are going to start popping up everywhere," he says, and the pace will accelerate heading into the midterms.
The government's relationship with the labs is a contradiction. Roetzer points to a telling example. Even as Anthropic gets treated like a regulatory villain in some Washington circles, its Claude Mythos model is reportedly being used by the National Security Agency, with Anthropic engineers embedded inside. "So who knows what is actually going on with the government and Anthropic, but it's this continuous dance," says Roetzer. "We're going to be really angry with them over here, and then we're going to use them over here to do this other stuff."
"Republicans might love that idea today, but if the Democrats come back in power, now they're going to hate the fact that the Democrats have a 50% stake. It always swings.
It might be a great idea when your political party is in power, but then you're going to hate that idea when the other party's in power."
— Paul Roetzer, founder and CEO of SmarterX, Episode 218 of The Artificial Intelligence Show
The speed of this shift is the real story. The idea of the federal government owning half of a major AI company would have been unthinkable six months ago. Now it is being discussed openly by both parties in the same week, which suggests both sides are reacting to the same thing: polling that shows the public is anxious about AI's impact on jobs and the economy. The precedent is already partly set, given the government's recent equity stake in Intel.
For business leaders, the practical takeaway is not to predict which proposal wins. It is to accept that AI policy is now volatile and bipartisan in a messy way, where today's good idea can flip with the next election. Anything built on the assumption of a stable, light-touch regulatory environment is built on sand. The smarter posture is to track the direction it's going — more government involvement rather than less — and plan for it rather than bet against it.
The White House meetings with AI companies. Whatever comes out of the discussions on a federal "partnership" will tell you how serious the equity idea really is or if it's simply political positioning ahead of the midterms. Will concrete dividend or ownership mechanics emerge or will it just stay talk?
State laws vs. federal preemption. The bipartisan House legislation would override state AI laws for three years, a move that could reshape the entire compliance landscape for any company deploying AI. How that preemption battle resolves will determine which rules businesses actually have to follow.
Only 13% of organizations report having the governance in place to scale AI responsibly, according to the 2026 State of AI for Business Report. That matters as Washington moves toward more aggressive oversight: The companies without internal governance are the least prepared for a world where AI rules tighten and shift with every election cycle.
The report is built on more than 2,100 responses from professionals across roles, functions, and industries, and it covers exactly where organizations stand on governance, adoption, training, and tooling. If you are trying to get ahead of the coming policy turbulence, it is the clearest benchmark for where you actually stand. Read the full report →