Is Government Involvement Slowing AI Progress? My Observations of What's Happening Now
June 20, 2026 | AI Trends
A purely observational piece, not a critique of any government, party, or policy.
Note: This one runs longer than my usual posts. There's a lot happening at once, and I didn't want to shortchange any of it. Grab a coffee.
The past two weeks have been unusually eventful at the intersection of artificial intelligence and government policy. In just ten days, the White House signed a new executive order, Congress released a sweeping 269-page AI bill, and two prominent legislators proposed an outright halt to new AI data center construction. For anyone watching AI development closely, the question is natural: is all of this activity changing the pace at which AI moves forward?
This post isn't here to argue whether any of these moves are good or bad, wise or misguided. That debate is important and worth having, just not here. What follows is simply my attempt to lay out what happened and observe, as clearly as possible, what effect it appears to be having on the speed of AI progress.
What Happened This Week
The White House Executive Order: A Voluntary 30-Day Review Window
On June 2, 2026, President Trump signed an executive order titled "Promoting Advanced Artificial Intelligence Innovation and Security." The headline provision: major AI companies are asked to voluntarily submit their most powerful new models for up to a 30-day government review window before releasing them to trusted partners.
A few things are worth noting about this:
- The word voluntary is doing a lot of work here. Companies are not legally required to participate.
- The original draft proposed a 90-day review period. Following pushback from the industry, that was walked back to 30 days.
- The order explicitly states that "nothing shall be construed to authorize the creation of a mandatory governmental licensing, preclearance or permitting requirement." In plain terms: no government permission slip is required to ship AI.
The practical effect, for now, appears to be modest. Companies that opt in may slow their external release timelines by up to a month for frontier models, but their internal development continues uninterrupted. If you measure AI progress by when the public gets access to new models, this could introduce a brief delay for some releases. If you measure it by when new models are trained, the needle barely moves.
Congress Floats the Great American AI Act
On June 4, Congressman Jay Obernolte (R-CA) and Congresswoman Lori Trahan (D-MA) released a 269-page bipartisan discussion draft called the Great American Artificial Intelligence Act of 2026.
This is big in scope. Key provisions include:
- Frontier model transparency requirements: companies with $500M+ in annual revenue that have trained a frontier model would face binding federal development obligations.
- State preemption: states would lose the ability to legislate on how AI systems are built, though they could still regulate how those systems are used.
- A three-year freeze on state AI laws targeting model development.
This is still a discussion draft, not yet a passed law. But it signals that comprehensive federal AI governance is no longer a distant possibility. It's being actively designed. Companies that were previously navigating a patchwork of state regulations now face the prospect of a single federal framework, which could mean more clarity, but also more compliance overhead, down the road.
The Anthropic Situation: Be Careful What You Wish For
Perhaps the most interesting subplot of the week involves Anthropic, the AI safety company behind Claude. For years, Anthropic has been one of the loudest voices in the industry calling for stronger government oversight of AI. The company donated $20 million to a political group backing AI regulation, and CEO Dario Amodei published an essay just days ago advocating for "serious and binding regulation of AI," explicitly including the government's ability to block models deemed unsafe.
Then the government blocked one of their models.
Following a federal export control order citing a "narrow potential jailbreak" that could allow foreign nationals access, Anthropic was forced to suspend its Fable and Mythos models. The company pushed back, arguing that applying this standard across the industry would "essentially halt all new model deployments." The Trump administration had separately designated Anthropic a "supply chain risk" over its pro-regulation stance, creating a complicated dynamic.
As CNBC put it plainly: Anthropic asked for regulation, and Washington went much further than expected. It's a genuinely interesting case study in how policy, once set in motion, rarely stays neatly within the boundaries anyone imagined for it.
Sanders & AOC Propose Moratorium
Perhaps the most striking legislative development came from Senator Bernie Sanders and Representative Alexandria Ocasio-Cortez, who proposed the AI Data Center Moratorium Act, a bill that would impose an immediate federal halt on constructing or upgrading data centers with a power demand of 20 megawatts or more, until "strong national safeguards" are in place.
To put that threshold in context: a single large AI training cluster can easily exceed 20MW. This proposal, if passed, would put a hard stop on the physical infrastructure that AI development depends on.
Most analysts and even some Democratic colleagues view this bill as unlikely to advance. Senator Mark Warner of Virginia called it "idiocy," arguing that pausing U.S. data center construction would hand an advantage directly to China and other competitors. Still, it represents a genuine strand of legislative thinking: that the pace of AI development may need to be deliberately slowed by constraining the compute it runs on.
So, Is AI Progress Actually Slowing Down?
Based on what we can observe right now: not significantly yet, but the direction of travel is toward more friction.
The executive order is voluntary and its impact on model development timelines is minimal. The Great American AI Act is a draft being circulated for comment, not a law. The moratorium bill faces an uphill battle in both chambers. Colorado, one of the most proactive states on AI regulation, is still in the process of implementing its own law.
What has changed is the environment in which AI companies operate. The hands-off era of AI oversight is clearly ending. Legal and compliance teams at AI companies are growing. Lobbyists are busier than ever. Executives are spending time in Washington that they are not spending on product. The Anthropic situation shows that even voluntary engagement with regulators can produce unexpected consequences. When companies start building with one eye on potential regulations, that influences what gets built and how fast.
None of that is the same as saying government involvement is stopping AI. It isn't, not remotely. But it is reasonable to observe that the overhead around AI development is rising, and that overhead has a cost measured in time and attention, even if it's hard to put a precise number on it.
Does This Help Anyone Else Catch Up?
This is the question that doesn't get asked enough, and the data makes it uncomfortable to ignore.
According to the Stanford AI Index 2026, China has narrowed the performance gap between its best AI models and America's best to just 2.7%. Three years ago that gap was between 17 and 31 percentage points. China did this while spending 23 times less on private AI investment than the United States. When DeepSeek-R1 briefly matched the top US model in early 2025, it was built at a fraction of the cost and compute. The efficiency story is real, and it's accelerating.
That context reframes the regulatory picture considerably. Any friction added to US frontier development, whether voluntary review windows, compliance overhead from the Great American AI Act, or the mere uncertainty of operating in a rapidly shifting regulatory environment, doesn't slow down the whole world. It slows down the companies subject to it. And the companies subject to it are almost entirely American.
This isn't a novel concern. Senator Warner's "idiocy" comment about the moratorium proposal was explicitly about China. The Carnegie Endowment for International Peace published a piece this month arguing that Trump's executive order is unlikely to stymie US competition with China, in part because the order is voluntary. That's reassuring for now. But the more binding proposals on the table carry a different risk profile.
There's also a domestic dimension worth noting. The Great American AI Act defines its obligations around companies with $500 million or more in annual revenue that have trained a frontier model. That threshold captures the big names: OpenAI, Google DeepMind, Meta, Anthropic. It largely exempts smaller domestic competitors. So regulation targeted at the frontier could, somewhat ironically, tilt the domestic playing field toward startups operating just below that threshold. Whether that's a feature or a bug probably depends on who you ask.
One more data point that deserves attention: the number of AI researchers choosing to move to the United States dropped 89 percent since 2017, and 80 percent in just the last year alone, according to Stanford's index. That trend predates the current regulatory wave, but it signals that the US talent advantage in AI is not guaranteed. Regulatory uncertainty is rarely the top reason researchers choose where to work, but it's rarely irrelevant either.
None of this means regulation is wrong. It means the effects of regulation don't stay neatly within borders, and a framework designed to slow the most powerful US labs down also affects how the US compares to everyone else. That's worth keeping in view.
The Bigger Picture
What's happening in Washington right now is, in many ways, an inevitable collision between a technology that moves faster than policy and institutions that are trying to catch up. Multiple competing visions are on the table simultaneously: the White House's voluntary framework, Congress's comprehensive federal bill, progressive proposals to slow infrastructure growth, and states trying to hold onto their own regulatory authority.
That's not unusual for a technology this consequential. The same collision happened with the internet, with social media, and with financial technology. In each case, the regulatory environment eventually settled into something, and in the meantime, the technology kept moving.
For now, my observation is simply this: the question isn't whether government will get more involved in AI. It already has. The more interesting question, one worth watching closely, is whether that involvement ultimately shapes AI into something safer and more trustworthy, or whether the added friction changes what gets built and by whom.
That story is still being written.
Sources: Trump Signs AI Executive Order Including 30-Day Review Period (Deadline). Trump's New AI Safety Order Seeks Voluntary Review of New Models (NPR). Trump Just Signed a Scaled-Back AI Executive Order (24/7 Wall St.). Anthropic Asked for Regulation. Washington Went Much Further (CNBC). Anthropic Incident Leaves Confusion About Trump Administration's AI Regulation (NPR). Unpacking the Great American AI Act (DLA Piper). Bipartisan AI Draft Proposes Three-Year Preemption of State Laws (Roll Call). Sanders, Ocasio-Cortez Announce AI Data Center Moratorium Act (Senator Bernie Sanders). Sanders and AOC Unveil Data Center Moratorium Bill (Axios). Stanford AI Index 2026: China Narrows US Lead to 2.7% While Spending 23x Less (The Next Web). Trump's AI Order Won't Stymie U.S. Competition with China (Carnegie Endowment for International Peace). Great American AI Act: What It Means for Companies (Captain Compliance).
Written by Travis Raveling, Founder PAID LLC, co-authored and edited by AI.
About PAID LLC: We help businesses understand, implement, and get ROI from AI tools and emerging technology. Learn more at paiddev.com/about.