AI Deepfakes Are Taking Over American Political Ads — And Nobody Can Stop Them

It wasn't. The "footage" showing Kentucky Rep. Thomas Massie, a staunchly conservative Republican, holding hands with Democratic Reps. Alexandria Ocasio-Cortez and Ilhan Omar as the supposed "throuple" checked into a hotel room together — was entirely fabricated using artificial intelligence. The ad, bankrolled by a super PAC supporting Massie's opponent ahead of the May 19 Republican primary, declared the invented affair "worse than adultery" and "a complete and total betrayal of President Trump."

Welcome to political advertising in 2026 — where none of it can be trusted.

A Flood of Fakes Hits the 2026 Midterms

The Massie ad is not an isolated incident. Across the country, the 2026 midterm elections have become a testing ground for AI-generated political content on a scale American democracy has never seen before. From congressional races in Kentucky to a Hollywood-flavored Los Angeles mayoral contest, deepfake technology has moved from a theoretical threat to an operational campaign weapon — and the political and legal guardrails meant to contain it are nowhere close to keeping up.

In the same Kentucky race, a separate PAC ran an AI-generated ad depicting Massie's opponent Ed Gallrein fleeing in terror from a Trump rally, then abandoning the former president on what appeared to be a World War II battlefield — none of which ever happened. Neither ad included the disclosures required under a Kentucky state law passed in 2025 to regulate synthetic media in political advertising.

In Maine, a PAC backing Republican Jonathan Bush in the gubernatorial race released an ad composed entirely of AI deepfakes of rival Robert Charles — including one fabricated scene in which Charles holds bags of money while a digitally recreated Barack Obama places a hand on his shoulder.

In Texas, the National Republican Senatorial Committee released an ad in which Democratic Senate nominee James Talarico appeared to speak directly into a camera for more than a minute — except it wasn't really him. The entire video was AI-generated. The words "AI GENERATED" appeared in small text in the corner. Texas law prohibits deepfake campaign ads only within 30 days of an election; the ad ran during the primary window, beyond the law's reach.

Spencer Pratt, Batman, and the Future of Political Branding

Not every AI political ad is designed to destroy. In Los Angeles, former reality television personality Spencer Pratt — a Republican running for mayor against Democratic incumbent Karen Bass — has become the unlikely face of AI's creative potential in politics.

A series of viral AI-generated videos, produced by filmmaker Charles Curran and circulated on X, depicted Pratt soaring through a burning, dystopian version of Los Angeles on a futuristic Star Wars-style vehicle. In the Gotham-inspired ad, AI versions of Bass and California Governor Gavin Newsom appeared as comic book villains delivering sinister lines about the city's decline. "The only thing that can stop us is someone telling the truth," the AI-generated Bass character says with a grin. "As long as they don't have any hope, the city's ours."

Pratt's campaign did not produce the ads — Curran created them independently as a supporter. But the videos generated enormous national attention, drawing praise from Jeb Bush ("maybe the best political ad of the year"), former Rep. Matt Gaetz ("a maximalist expression of what political ads can do"), and Fox News host Laura Ingraham. Bass called the ads "a very dangerous trend" and expressed concern that the violent imagery could provoke unstable viewers.

Pratt himself, who lost his Pacific Palisades home in the January 2025 wildfires and announced his candidacy on the one-year anniversary of the disaster, has channeled the AI buzz into a dark-horse campaign that has rattled Los Angeles's political establishment. He launched his bid as a registered Republican but emphasized the race is officially nonpartisan, saying "there will be no D or R next to my name" on the ballot.

Democrats Are Also Using AI — Just More Quietly

The avalanche of AI ads has come predominantly from Republican-aligned groups, but Democrats have not stayed on the sidelines. A Democratic House challenger in upstate New York released an AI-generated attack video targeting Rep. Elise Stefanik last year. Rep. Jasmine Crockett of Texas faced accusations of using AI to artificially inflate crowd sizes in one of her campaign ads. The line between digital enhancement and outright fabrication has become increasingly difficult to draw.

"The use of AI in election ads has exploded in the 2026 midterms, with some campaigns using the new technology to push the limits of negative campaigning," Axios reported in May, noting that regulation efforts remain stalled at the federal level.

The Laws Are Not Working

Roughly half of U.S. states have passed some form of legislation addressing AI in political advertising, but the patchwork of laws contains significant holes. Most require only disclosure — a small "AI GENERATED" label — rather than outright prohibition. Many apply only in windows close to election day, leaving primaries largely unprotected. Almost none include criminal penalties severe enough to deter well-funded super PACs.

Kentucky passed bipartisan legislation in 2025 requiring clear disclosure of AI-generated content in political ads. The law allows candidates to sue for an injunction if the disclosure is absent — but it does not ban the deepfakes themselves. When the anti-Massie "throuple" ad ran without proper disclosure, Massie called it "disgusting and defamatory." Former Rep. Marjorie Taylor Greene argued the ad violated the federal Take It Down Act, which prohibits non-consensual intimate imagery — though legal experts questioned whether the definition stretched that far.

Rep. Joe Morelle, the top Democrat on the House Administration Committee, told Axios that regulating AI on the campaign trail will "absolutely" be a congressional priority — but that action remains in the future tense, while the ads are running right now.

Why They Work — Especially on Older Voters

The scientific case for why AI political ads are so potent is not complicated: the human brain is not wired to scrutinize video footage the way it scrutinizes written claims. Decades of psychological research show that visual information — especially footage that mimics the grainy, unpolished aesthetic of "real" leaked or surveillance video — triggers faster, less skeptical processing than text.

Older voters, who represent the most reliable election turnout demographic and who grew up in an era when video footage was inherently trustworthy, are particularly susceptible. "There's going to be a period where we're going to be influenced by what AI presents," one San Jose resident told Fox News, speaking about AI's role in elections. "Public opinion will be shaped in a large way."

The Massie throuple ad was engineered to look exactly like leaked surveillance footage — a format that carries an implicit credibility. It doesn't announce itself as advertising. It presents itself as evidence.

What This Means for 2028

If 2026 is the year AI political ads went mainstream, 2028 is the year they will be industrialized. The presidential race — the most expensive, most watched, most contested political event in the world — will be the ultimate testing ground for everything being piloted in Kentucky congressional races and Los Angeles mayoral contests today.

The candidates who will compete for the presidency in 2028 — names like JD Vance, Marco Rubio, Gavin Newsom, and Josh Shapiro — will face AI-generated content about themselves at a volume and sophistication that has no precedent. Their faces, voices, and mannerisms are already extensively documented in public video — exactly the raw material AI image generation tools require.

The fundamental question that the 2026 midterms have raised — and left unanswered — is whether American democracy can adapt its norms of evidence and trust fast enough to survive the technology. When any scene can be fabricated, when any politician can be placed in any situation, and when the fabrications can be produced and distributed within hours of a news cycle, the epistemological foundations of political advertising collapse.

Voters in 2028 will need new habits of skepticism that most of them have not yet developed. Campaigns will need rapid-response fact-checking infrastructure built for video, not just text. Platforms will need detection and labeling systems that work at scale. Legislators will need to act before the election, not after it.

The 2026 midterms are the warning shot. The question now is whether anyone is listening.

 

Follow TalkBea for daily 2028 election coverage. See our Candidate Profiles and Live Poll Tracker for the latest on the presidential race.