Mind the Gap: Democracy and Data Centers

John Pitts is the co-founder of Open Banker. In his spare time he works at Affirm as the Vice President of Public Policy, Public Affairs, and Social Impact. Prior to Affirm, John was Head of Industry Relations and Head of Global Policy at Plaid. Prior to Plaid, John was the Deputy Assistant Director for Intergovernmental Affairs at the Consumer Financial Protection Bureau.

Hot take: the Black Death was actually kind of great. For centuries, the feudal peasant class were essentially tied to the land they were born on, earning the same wages as their great-grandparents. The aristocracy owned the land, everyone else worked it, and the supply and demand curve for agricultural labor sat in a steady — if Malthusian — equilibrium. But while the plague halved labor supply, it had no impact on labor demand, and those surviving workers suddenly found themselves in a position to demand substantial wage increases. 

The aristocracy… did not care for this. The Crown and Parliament responded with the Ordinance of Labourers and the Statute of Labourers, two laws that attempted a Canute-like command for the economic tide to ebb. They fixed wages at pre-plague levels and restricted laborers from seeking better employers.¹ The result was a sudden and massive gap between what the population wanted and government policy. And, more specifically, the result was the 1381 Peasants’ Revolt, when thousands of English workers marched on London, burned government records, and forced the King to meet them face-to-face and grant concessions. While they eventually lost — the King reneged, rebel leaders were killed and their followers hunted down, and Spiderman made a movie about it — the lesson still holds. If the gap between government policy and the desires of the governed gets too big, too fast, the governed will make themselves heard.

The Butlerian Prequel

The biggest gap between the desire of the governed and the policies of the government in the US today is AI policy. The resistance to data centers is the evidence, for those who have eyes to see it. 

AI was certainly a concept during the 2024 election. ChatGPT had already become a household name, though progress on AI models has been so swift that it is easy to forget that the 2024 version of LLMs was closer to a party trick with promise than the GDP juggernaut we see today. That party trick wasn’t a political issue of any significance. Inflation, ongoing dissatisfaction with the Covid response, and cultural issues all mogged AI policy like they were ASU frat leaders.

But election issues seldom translate cleanly to administration priorities. Hyperscalers and their investors convinced the President that US AI dominance was the geopolitical imperative of the century, a bigger priority than any culture war.² And Silicon Valley, stung by the Biden administration’s relatively aggressive posture to regulating their innovations during the prior presidency, viewed domestic regulatory barriers as the primary enemy — even more so than China. The result was the Trump administration’s AI Action Plan, which made American AI dominance an explicit national objective. A July 2025 executive order struck the first blow against lugubrious process by accelerating the most critical AI infrastructure. It directed agencies to accelerate data-center development by easing federal regulatory burdens, expediting environmental review, and fast-track financial support for qualifying projects. 

Even as late as 2025, this was not a repudiation of a well-formed public consensus against AI. It was a fast plan for building faster, up there with Operation Warp Speed as an example of an uncharacteristically fast development and deployment of policy by the federal government. Unfortunately for the administration, AI technology moved even faster. And public opinion moved with, and decisively against, the thinking machines.

That is the sound of a gap opening. Fortunately, US democracy is a better designed government than feudal England’s. Gaps between the will of the governed and the plans of the federal government open all the time, but with Representatives elected every two years, Presidents every four, and Senators every six, there are ample opportunities for voters to petition for a shift in federal policy. It is messy and imperfect, but it generally keeps things within a standard deviation of most people’s desires — close enough that people may grumble but they’ll usually just wait for the next election.

Unfortunately it isn’t always fast enough, as what we’re seeing right now demonstrates. AI opposition swelled rapidly after 2024, at the same time that the federal government was making a tremendous bet on AI support. At the same time, AI adoption was more rapid than any technology in human history, and that technology was iterating rapidly, making the impact on everyone’s lives both obvious and — for most people — worryingly exponential. 

With no real levers to change federal policy, AI opponents identified the same infrastructure chokepoint the Administration had and began to flood that Tocquevillian bastion of American politics: local zoning hearings. In Chesapeake, Virginia, residents organized within weeks, filled a council meeting, spoke for more than two hours, and won a unanimous vote blocking a proposed facility near homes. In Monroe County, Georgia, hundreds of opponents in matching “Say No To Rezoning” shirts helped persuade a zoning board to reject the rezoning needed for a proposed 900-acre project. Two young women in Utah smacked the bejeesus out of a Canadian billionaire in one of the funniest videos I’ve ever seen. Who needs White House access and multimillion dollar lobbying coffers when you have folding chairs in a municipal basement and the right to wait in line to speak at a microphone?

Zoning is one of the few places where an ordinary person can still put their body in the room where a consequential decision is being made. A federal AI strategy may feel untouchable. The comprehensive plan is not. Don’t believe me? Believe other John.

Birds of a Feather Vandalize Together

Local governments made a similar choice with Flock Safety’s automated license-plate readers. Police departments and city officials often describe the cameras as practical investigative tools:  they can identify stolen vehicles, locate missing people, and connect a car to a serious crime. San Diego, for example, reported that its system supported hundreds of investigations, arrests, and stolen-property recoveries. Those are real public benefits. My street has 3 of the 67 Flock cameras in all of Washington, DC, but as we’re experiencing one or two drive-by shootings every year I’m not mad about it.

I am, however, mad about cops using Flock to stalk their ex-wifes and girlfriends. Which they are doing. A lot. I’m also annoyed that Congress has failed over, and over, and over again to pass a federal privacy law so that I have a basic set of rights on how the massive amounts of data collected about me will and won’t be used. That annoyance didn’t matter until recently. The grim joke in DC policy circles is that Americans may say they want privacy protections on their data, but their revealed preference is to smash the “accept all cookies” or any other data use button as fast as they can so that they can start using the product or service they want.

That’s no longer the case with Flock. Opposition has moved outside lawful democratic channels. Cameras have been painted, struck, and cut down. There are hours of YouTube videos instructing people how to destroy them without getting caught (these videos are, of course, brought to you by data centers). Why? It’s not like we haven’t had security cameras before. If I had to guess, it is that Flock has done a great job of emphasizing how deeply embedded AI is into its technology. It is one thing to worry about a security camera when the only one watching the video is Paul Blart. It is quite another when powerful AIs capable of crunching trillions of bits of data are looking at it — and combining it with all that other data of yours you suddenly realize isn’t subject to any meaningful protections.

Local governments have an advantage on the Flock issue because, like zoning boards, they are close to the people and fast to react. The Associated Press reports that more than 50 agencies or jurisdictions canceled, suspended, rejected, or deactivated Flock systems in 2026. But not everyone holding a saw thinks that they have meaningful access to City Hall. As in 1381, the destruction over government property is likely to continue for as long as the desire for new privacy protections in an AI world diverges from Congress’ sclerotic inability to even hover its mouse over GDPR and hit cut and paste.

Hey, So This is a Financial Services Regulation Newsletter?

The Peasants’ Revolt did not occur because medieval workers had developed a sophisticated theory of democratic representation. They had not. It occurred because the world had changed, ordinary people understood that the change was dramatically impacting their lives, and the government used its power to drive policy in the opposite direction of their desires. With formal institutions unresponsive to their wants, people reached for other tools. 

Like it or not, the most formal institution in the US — behind the government — is the financial services system. A banking system that has operated for 27 years on the content assumption that Gram-Leach-Bliley is the only privacy law that will ever apply to it.³ A banking system that is rushing to incorporate AI into its systems, lest they wake up one day to find that Grok has X Moneyed all of their deposits away. A banking system that was shaken enough by the public anger of 2008 to cut deals to bring the Consumer Financial Protection Bureau into existence, but is now far enough away from Zuccotti park that it feels safe advocating for the agency to be shrunk to the size where it can be drowned in a bathtub. A banking system that I am part of and want to do good, as well. 

All of the attention is focused on government and hyperscalers. For now. But the things financial institutions are saying about how they can use AI to anticipate exactly when you might be interested in a loan offer to fund your next Kalshi prediction, find the highest auto loan rate you’ll still say yes to, or replace call center workers with LLMs, or engage in buying, selling, and investing on behalf of the consumer are being heard today. And remembered. Banks have a free front row seat to the brutal gauntlet data centers and Flock cameras must now run. They should not be eager to follow them. They should instead be applying the lessons of this blowback to their own plans and adopt policies consumers will find reasonable regardless of how permissive their current regulator is willing to be. They should read the Consumer Reports AI standards and the Colorado Attorney General’s regulation on consequential decisions by AIML. And they should read them with urgency. Because if the gap between bank AI use and the demands of the public grows too large, things will, as the poet said, snap back to reality.

The opinions shared in this article are the author’s own and do not reflect the views of any organization they are affiliated with.

^[1]  Fun fact:  they were also the first vagrancy law, and made unemployment and idleness illegal as a way of preventing workers from refusing to work unless they got higher wages. Something to keep in mind if you want to look at current debates around homelessness laws, work requirements for Medicaid, and other social safety net programs through a Marxist dialectical lens. Which I don’t. But come on fellow capitalists, let’s maybe not give the other team uncontested layups?

^[2]  This isn’t an argument that they were or are wrong. FWIW, I’m inclined to agree with them.

^[3]  They aren’t wrong. So far. Every state privacy law has included a carve out for institutions that are subject to GLBA. Let’s check back on that in five years.

Open Banker curates and shares policy perspectives in the evolving landscape of financial services for free.

If an idea matters, you’ll find it here. If you find an idea here, it matters. 

Interested in contributing to Open Banker? Send us an email at [email protected].