The 2026 midterm elections are quietly becoming the first true artificial intelligence election, and most voters have no idea how much of what they are seeing was never real to begin with.
According to the Wesleyan Media Project, at least 164 political ads created or enhanced with AI had been identified by early September 2026, accounting for nearly $80 million in ad spending. The project cautions the true number is likely far higher, since AI use can be difficult to confirm.
How Big Is AI Spending in the 2026 Midterms?
AI generated advertising has moved from novelty to standard practice. Wesleyan found roughly 7 in 10 AI ads carried no disclosure of the technology, and a New York Times analysis found super PACs tied to OpenAI and Anthropic have spent $55.7 million influencing key 2026 races.
The Times examined Federal Election Commission filings as of late September and reviewed television and digital ads gathered by the tracking firm AdImpact. The finding reveals that the two leading AI developers are not just shaping the technology, they are shaping the political environment in which it will be regulated.
Meanwhile, Campaigns and Elections reports that 717 campaigns, from U.S. Senate races down to county commissions, ran ads on Meta and Google mentioning AI data centers alone. Of those, 342 were Democratic campaigns, 236 were Republican, and 58 were independent, third party, or nonpartisan.
What Do AI Attack Ads Actually Look Like?
Some are absurd by design. Others are engineered to deceive. As reported by WTOP and News4Jax, the campaign for North Carolina Republican state Sen. Michael Lee posted an ad in August attacking Democrat Jessica Bichler, a yoga instructor, showing her contorting into impossible positions with a disclaimer reading, "You guessed it! AI was definitely used to generate these silly video clips."
The Washington Post highlights the razor's edge Ohio Senate race, where Republican Sen. Jon Husted is defending his seat against Democrat Sherrod Brown, as a marquee contest where AI creative is transforming how campaigns energize voters. Texas Lt. Gov. Dan Patrick also released an ad using an AI generated image of his opponent, Democrat Vikki Goodwin.
The darker side is documented by The Guardian, which reported on a fake image of a Georgia mayor that went viral, and by The New York Times, which found Black voters are facing artificially generated fakes, character assassination campaigns, and tactics with names like "crisis profiteering" and "cynicism bombing" ahead of November.
"In 2026, not only are we going to see all of this at a velocity that no previous generation has faced," said Esosa Osa, founder of the nonpartisan nonprofit Onyx Impact, as quoted by The New York Times. "It's going to need to be community driven."
Are State Disclosure Laws Working?
So far, the evidence says no. A patchwork of state laws has emerged governing AI transparency in campaign ads, but reporting by WTOP shows candidates at times disregarding those laws, with 69% of identified AI ads carrying no disclosure.
The Conversation examined Louisiana's June 2026 law, which requires a clear warning on AI ads meeting certain conditions, including printed materials, online advertisements, and broadcasts. Even so, the analysis of two state laws found sharp limits in what legislators can actually accomplish as deepfakes go viral.
Texas offers another example. Houston Public Media notes that a 2019 Texas law makes fabricating a deceptive video with intent to influence an election a crime, yet the Paxton and Talarico Senate contest is on course to demolish the Texas record for campaign spending, with AI creative firmly in the mix.
Experts quoted across the reporting agree on one point: it is too soon to measure how AI ads move voters, with or without labels, but more effective regulation is imperative.
What Does This Mean for Campaigns and Voter Outreach?
For campaign professionals, the lesson is twofold. First, AI creative is now table stakes, offering what Virginia Tech's expert analysis describes as high quality, low cost promotions that gain voter attention. Second, trust is becoming the scarcest resource in politics, and campaigns that abuse it may face backlash.
Virginia Tech also notes that many voters now get news from AI chatbots, and the 2026 midterms may produce the first evidence of chatbots as voter influencers. That makes direct voter contact, the kind delivered through live conversations via HyperPhonebank, more valuable than ever. When voters cannot trust what they see on a screen, a real voice on the phone cuts through.
The same logic applies to message testing. Campaigns relying on HyperPulse synthetic polling can model how voters respond to AI generated content before committing millions to creative that might trigger a credibility crisis.
Internationally, the warning signs are flashing. EDMO documented a staged video during Berlin's September 2026 parliamentary election falsely claiming postal ballots for the AfD were being destroyed, part of a wider pattern of false postal voting claims across German state elections this autumn. South Africa's IEC Chairperson Mosotho Moepya has warned of "ward specific deceptions" driven by generative AI and hyper local campaigning, per Polity and News24.
How Should Campaigns Prepare for What Comes Next?
Smart campaigns are treating AI as both an opportunity and a threat surface. The opportunity is cheaper, faster creative. The threat is a hostile information environment where opponents, foreign actors, and anonymous super PACs can manufacture reality at scale.
The countermeasure is verified, direct communication. Robust voter outreach services, including phone banking, text programs, and field operations, give campaigns a channel that deepfakes cannot easily counterfeit. Campaigns can explore the full picture through our campaign strategy coverage as the cycle unfolds.
The Wesleyan data makes one thing clear: disclosure norms are being ignored in real time, and enforcement is lagging. Campaigns that build trust through authentic contact now will hold an advantage when, not if, the first major AI scandal of this cycle breaks before November.