Artificial intelligence is no longer just a campaign talking point in the 2026 midterms. It is becoming campaign infrastructure, campaign money, local zoning politics, and campaign media all at once.
The clearest signal is money. Transformer’s AI Campaign Finance Tracker, which says it is based on Federal Election Commission data and manually reviewed updates, showed $50.79 million in AI-focused super PAC spending as of June 28, 2026 at 02:16 UTC. The tracker splits that spending between $22.89 million from groups it classifies as “pro-innovation” and $27.91 million from groups it classifies as “pro-safety.”
That is not yet presidential-cycle money, but it is already enough to reshape individual primaries. New York’s 12th Congressional District became the first high-profile test case after state Assembly member Alex Bores, who coauthored New York’s RAISE Act for frontier AI safety disclosures, was targeted by AI-linked outside spending. The Verge reported that AI companies and aligned super PACs spent $27.41 million around the race, with $19.26 million from pro-Bores groups and $8.15 million from Leading the Future opposing him. Bores narrowly lost to Micah Lasher, 39.1% to 35%, but the race showed that AI policy can attract national spending even when local politics still decide the result.
The AI fight is not only about model rules
The emerging campaign issue is broader than whether Congress or states should regulate frontier models. It touches three things voters can see: who is paying for AI-friendly candidates, whether new data centers raise local costs, and whether campaign ads can still be trusted when synthetic media is cheap.
The data-center fight may be the most visible of the three. Axios reported on polling from Milltown Partners that found nearly half of 6,872 registered voters supported a temporary moratorium on new data-center construction, while 38% supported having a data center built near their home and 34% opposed it. Only 8% of respondents who opposed data centers said they knew of one near their home, which suggests the backlash is not simply a neighborhood-not-in-my-backyard reaction.
Other polling points in the same direction. Gallup found in a March survey that 70% of U.S. adults opposed building AI data centers in their local area, including 48% who were strongly opposed. The most common reasons were resource concerns, including water use, energy demand, grid constraints, environmental impact, and quality-of-life effects.
That matters politically because the next wave of infrastructure is not confined to the same places that already host most data centers. Pew Research Center found that 67% of planned U.S. data centers are in rural areas, while 87% of existing facilities are in urban areas. Pew also found that 39% of planned projects are in counties that currently do not have a data center. That shifts the politics from abstract AI anxiety to school tax breaks, utility bills, farmland, power lines, water, and local permitting meetings.
AI money is following policy chokepoints
The super PAC fight is moving through the same geography. The Guardian reported that AI-focused super PACs had raised more than $100 million this cycle and had spent $49 million across dozens of congressional races by June 22. The same report noted that much of the spending had flowed into NY-12, but also into races in Utah, Texas, Ohio, Georgia, and Kentucky, including places tied to data-center expansion and AI legislation.
The split is not simply “tech versus anti-tech.” The major camps are arguing over which rules should govern AI, who gets to write them, and whether state laws should be preempted by a federal framework. Leading the Future has positioned itself around a pro-AI, federal-rule approach. Public First and related groups have backed candidates and messages that favor stronger guardrails, even though some of their funding also comes from inside the AI industry.
That makes AI politics unusual. Companies are not only lobbying against regulation from the outside. Rival labs, executives, investors, labor groups, and policy networks are fighting over the shape of regulation from inside the same industry. The result is a campaign-finance map that looks less like a single tech lobby and more like a fight over whose AI worldview becomes law.
Synthetic ads are testing campaign norms before law catches up
The third pressure point is campaign media itself. AI-generated political ads have moved from novelty to routine attack format, and disclosure rules have not kept pace. Axios documented examples across Texas, Kentucky, Georgia, New York City, and Maryland, including ads that depicted candidates saying or doing things they did not actually do.
Some of those ads are obviously exaggerated. Others use synthetic video or images to make real claims feel more vivid, more humiliating, or harder to separate from ordinary campaign editing. That distinction matters because political advertising has always used caricature and selective editing, but generative AI lowers the cost of manufacturing scenes that never happened.
For voters, the practical problem is not only whether a fake video fools them once. It is whether campaigns can flood local races with synthetic material faster than journalists, platforms, campaigns, and election officials can verify it. The risk is highest in lower-profile races, where opposition research, local news capacity, and voter attention are all thinner than in presidential contests.
What to watch next
The strongest sign that AI has become a durable election issue will not be one viral deepfake or one expensive primary. It will be whether the same pattern repeats across state and congressional races: outside AI money targets candidates tied to regulation, local data-center fights become campaign material, and synthetic ads become standard tools for shaping voter impressions.
The next useful questions are concrete. Which candidates receive AI-linked PAC support after filing deadlines? Which data-center projects become local campaign issues? Do campaigns disclose synthetic media voluntarily, or wait for legal requirements? Do platforms label political AI content consistently across paid ads, short videos, and reposted clips? And do voters punish campaigns for using synthetic scenes, or treat them as another form of attack-ad theater?
The 2026 midterms will not be decided by AI alone. Inflation, foreign policy, party control, abortion, immigration, and local candidate quality will still matter more in many races. But AI has crossed an important threshold: it is now part of the machinery of elections, the physical infrastructure debate in communities, and the rules candidates are fighting over before Congress writes the next major technology law.
