The Violent AI Backlash: When Cities Push Back

From city council chambers to residential streets, communities across the United States are mounting organised resistance to AI infrastructure and in some cases, that resistance is turning dangerous.

The promise of AI-powered smart cities has long been framed as a top-down proposition: technology companies build the infrastructure, governments adopt the tools, and citizens reap the benefits. But a growing number of communities are rejecting that model, loudly, democratically and in a handful of extreme cases, violently.

In April 2026, an Indianapolis city councillor who voted to approve rezoning for a data center returned home to find 13 bullet holes in the exterior of his house and a note under the doormat reading "NO DATA CENTERS." Days later, a man allegedly threw a molotov cocktail at OpenAI CEO Sam Altman's home, then headed to OpenAI's headquarters where he allegedly threatened to burn the building down. He has since pleaded not guilty to charges including attempted murder. These remain isolated incidents, but researchers who study political violence say the structural conditions that produce such acts are widening.

The Urban Flashpoint: Data Centers

At the centre of the conflict is physical infrastructure. AI systems require enormous data centres, large facilities with significant energy and water demands, and they are frequently sited in or near residential communities with minimal public consultation.

The scale of organised resistance is now well documented. Energy analyst Robert Bryce tracked at least 70 US communities that imposed restrictions on or rejected data centre projects between 2021 and early 2026. A record number of planned projects were cancelled in Q1 2026 alone. Maine became the first state to pass a statewide data centre moratorium, later vetoed by the governor and at least 12 other states have introduced similar legislation.

Community objections are grounded in concrete urban concerns: rising local energy costs, excessive water consumption, noise, limited permanent job creation, and the use of public tax incentives for projects that primarily benefit large corporations. In Port Washington, Wisconsin, residents passed what is reportedly the first referendum of its kind in the United States, requiring public approval before tax incentives can be granted to major projects including data centres. In Festus, Missouri, voters removed every city councillor who had approved a $6 billion data center project.

A community guide called "How to Stop a Data Center," produced by a group in Michigan, notes that demonstrating outside local officials' homes has become an effective organising tactic, a measure of how far the conflict has moved from formal governance channels into direct action.

Investment banks have taken note. Morgan Stanley warned clients that "public pushback is emerging as a binding constraint" on AI infrastructure expansion. Jefferies told investors that community-level battles were "sapping confidence" in the sector.

A Governance Vacuum at the Local Level

What the backlash reveals, above all, is a governance gap. Decisions about where data centres are built, how much energy they consume, and what public subsidies they receive have largely been made without meaningful community input. The result is that local elected officials, city councillors, planning commissioners, zoning board members, are absorbing the anger that communities feel toward decisions made far above their level.

Yannick Veilleux-Lepage, a researcher who studies technology and terrorism, wrote last month that AI "generates the structural conditions historically associated with the onset of political violence". Local officials, he noted, are in a particularly exposed position: "Where national figures are unreachable, local policymakers who approved the data centre become the proxies for the same structural anger." Following the shooting in Indianapolis, the city council introduced a measure that would allow officials to keep their home addresses private.

The Soufan Center, a nonpartisan security research organisation, has documented a rise in direct threats against individuals, policymakers, and corporations involved with AI. The most common threats involve physical sabotage of data centres — facilities that sit, in many cases, within or adjacent to the communities they are inflaming.

What Citizens Are Telling Pollsters

The anger is not difficult to quantify. The 2026 Stanford AI Index found that only 38% of Americans now view AI positively, the lowest figure the report has ever recorded. A Gallup survey found that just 18% of young Americans feel hopeful about AI. Over 70% of all Americans believe the technology is moving too fast, per Economist/YouGov, a view shared by 68% of Republicans and 77% of Democrats.

The anxiety crosses political lines in ways that are unusual in the current environment. Bernie Sanders recently wrote that "AI oligarchs do not want to just replace specific jobs. They want to replace workers." Steve Bannon, Trump's former chief strategist, made near-identical remarks. As The Atlantic observed, this emergent "Bernie-to-Bannon" coalition reflects a genuinely bipartisan concern – one rooted less in ideology than in a shared scepticism about who benefits when large technology companies reshape local communities.

Stanford law professor and AI expert Nathaniel Persily framed the underlying dynamic clearly: "Disruption has winners and losers. For many Americans, they're not convinced they're going to be the winners, and they base that conclusion on the history of technology over the last 20 years."

Among Americans earning under $50,000 a year, 56% fear AI will replace jobs their families depend on. One poll found that the group most optimistic about AI's impact on their daily lives are households earning over $200,000 a year. The gap between who bears the disruption and who captures the gains is not merely a perception issue, it shapes how communities respond to infrastructure proposals arriving in their neighbourhoods.

The Industry's Response

The AI industry is investing in public goodwill alongside physical infrastructure. Anthropic committed $200 million alongside the Gates Foundation for programmes in health, education, and agriculture in lower-income countries. OpenAI struck a deal with Malta to provide every citizen with free ChatGPT Plus access following an AI literacy course. George Osborne, head of OpenAI's countries initiative, described the direction: "Intelligence is becoming a national utility and all governments have an important role to play."

On the energy side, one of the most concrete local concerns is significant capital is flowing toward cleaner solutions. Geothermal company Fervo Energy surged 33% on its IPO debut, crossing a $10 billion market cap. Nuclear startup X-energy raised $1 billion in an oversubscribed offering. Both pitch themselves as clean baseload power providers specifically for AI data centre demand.

But as The Atlantic noted in its recent analysis of the backlash, "if the tech industry truly believes that a simple change in messaging will quell the backlash, then they are misunderstanding the problem entirely." Goodwill investments and narrative adjustments do not resolve the underlying governance question: on whose terms, with what community input, and with what local benefit does AI infrastructure get built?

What Cities Need Now

The current moment points toward a need for clearer frameworks governing how AI infrastructure is sited, consulted on, and integrated into existing urban planning processes. The communities that have pushed back most effectively, through referendums, council votes, and state-level legislation, have done so by using existing democratic tools. The challenge is that those tools were not designed with the speed or scale of AI infrastructure rollout in mind.

What is clear is that the smart city vision, in which technology enhances urban life, improves services, and strengthens communities, cannot be delivered over the sustained objection of the communities it is meant to serve. The backlash will not stop AI development. But it will shape where it goes, how fast it moves, and whether it arrives with or without public consent.

That question is now being answered, city by city, council meeting by council meeting and in a small number of cases, on residential streets in the dark.

Sources: