# AI-Generated Deepfake Attack Ads Spread Across Early Voting States
As South Carolina opens early voting on Tuesday for its state primaries, voters scrolling their phones and flipping through TV channels will encounter a political landscape fundamentally altered by artificial intelligence. Deepfake attack ads depicting candidates in entirely fabricated scenarios have proliferated across contested races nationwide, and the regulatory framework to stop them remains almost nonexistent.
The AI wars have arrived just in time for election season, and nobody is fully prepared.
Fake Candidates, Real Consequences
In South Carolina alone, AI-generated ads have already targeted some of the state\u2019s most prominent officials. One ad depicts Attorney General Alan Wilson ignoring a phone call from President Donald Trump. Another shows Lt. Gov. Pamela Evette hooked up to a lie detector test. Neither scenario ever happened. Both were created entirely with generative AI tools that can now produce photorealistic video in minutes.
The trend extends well beyond the Palmetto State. In March, the National Republican Senatorial Committee released what experts have called the most realistic political deepfake used in a U.S. campaign to date: a fabricated video of James Talarico, the Democratic nominee for a U.S. Senate seat in Texas, speaking directly into the camera for over a minute. The ad ran without legal consequence.
In Kentucky, a MAGA-aligned PAC deployed an AI-generated video showing Republican Congressman Thomas Massie holding hands with Democratic Representatives Alexandria Ocasio-Cortez and Ilhan Omar. A separate ad depicted GOP congressional candidate Ed Gallrein fleeing from President Trump on a fictional battlefield. In the Texas Senate race, dueling deepfakes emerged from both sides: one showed Attorney General Ken Paxton riding in a car with women who are not his wife, while another fabricated a scene of Senator John Cornyn dancing with a Democratic congresswoman.
At least 15 campaign ads featuring AI-generated content have run since November 2025, according to a Reuters investigation, and the pace is accelerating as primary season heats up.
The Detection Problem
Alon Yamin, CEO and co-founder of Copyleaks, an AI detection software company, told Live 5 News that the technology is outrunning the average voter\u2019s ability to spot fakes.
\u201cIt\u2019s becoming harder and harder to distinguish between AI and human content,\u201d Yamin said. \u201cThe bar for fraud is just so low. Anyone can create whatever they want with AI at this point, and that creates a serious problem for the political landscape.\u201d
Yamin offered practical guidance for voters: watch for smudging around hair and hands, check whether lip movements match spoken words, and be skeptical of low-quality or blurry video, which can mask AI artifacts. But he acknowledged these telltale signs are disappearing as the technology improves.
\u201cUnless you\u2019re using an AI detection technology that could look at the frames and the pixels and tell you with a high level of accuracy if this is AI or not, you\u2019re not able to really know,\u201d he said.
A 2025 peer-reviewed study published in the Journal of Creative Communications reinforced that concern, finding that viewers struggle to identify deepfake videos and that their political opinions are measurably affected by the misinformation they contain, with impressions persisting even after corrections are issued.
A Regulatory Vacuum
The federal regulatory response has been fractured and slow. The Federal Election Commission issued an interpretive rule clarifying that existing prohibitions on fraudulent misrepresentation apply regardless of the technology used, but it has stopped short of creating AI-specific regulations, instead evaluating complaints case by case. The Federal Communications Commission has banned AI-generated voices in robocalls but has not extended that prohibition to digital ads, television spots, or social media content.
That leaves a patchwork of state laws as the primary guardrail. Twenty-eight states have passed legislation addressing AI in political advertising, but most focus on disclosure requirements rather than outright bans, and few have been tested in court. South Carolina\u2019s own deepfake laws primarily target explicit sexual content, not political manipulation.
\u201cWe\u2019re seeing different regulations in different states,\u201d Yamin said. \u201cI think really what\u2019s missing is to have something on a federal level where all states are working in the same landscape.\u201d
Republicans have utilized the technology more frequently than Democrats this cycle, according to politics experts and a Reuters review of publicly available ads, though both parties have access to the same tools and the gap may narrow as November approaches.
What to Watch
The 2026 midterms are shaping up as the first major U.S. election where AI-generated content is not an edge case but a mainstream campaign tactic. With 30 states now holding some form of deepfake legislation and Congress showing no signs of imminent action, the burden of verification falls increasingly on voters themselves.
As early voting opens in South Carolina and primary season accelerates nationwide, the central question is no longer whether deepfake political ads will influence elections. It is whether anyone, from regulators to platforms to individual voters, can keep up with a technology that evolves faster than the rules designed to govern it.
The wild west of AI-generated political content is here. The sheriff has not yet arrived.