Political campaigns are spending more money on artificial intelligence than ever before. The technology is lowering the cost of producing attack ads, memes, and other targeted messages, giving candidates who can afford it a new advantage. That advantage may not last.
AdImpact identified 29 campaign ads in federal and gubernatorial contests that disclosed AI content as of two weeks ago. The Wesleyan Media Project found at least 164 AI-influenced ads this cycle, counting ads that did not disclose AI use. The difference matters. Disclosure is voluntary, and campaigns that do not admit using AI are still using it.
Jesse Hunt, a veteran Republican consultant, told reporters that AI “certainly allows you to produce more content in varying styles and more cost effective.” Chris Moyer, a Democratic strategist, agreed, saying AI “gives campaigns more tools for less cost” and “democratizes tools that campaigns used to have to spend valuable resources on.”
How AI Changed One Agency’s Workflow
Will Ritter runs Poolhouse, a Republican ad shop. He described the shift bluntly: “We went from last cycle doing basically doing nothing with AI to most of our ads now have some AI component.” The change was fast, and it changed what campaigns expect from their shops.
Ritter added that “You still need to able to know what you’re doing with the tools” and that “we have professionals who know how to use and integrate them…to communicate a message, not to show off cool visual tech.” The message is clear: AI is a tool, not a magic wand. Campaigns that treat it as the latter are wasting their money.
Ciro Riccardi is a campaign adviser for Republican Rep. Mike Lawler of New York. His view is direct: “Utilize[s] every legitimate tool that helps our team improve productivity,” he said. “Any campaign that doesn’t would be foolish.” Lawler’s campaign has reported spending on AI technology vendors in FEC disclosure reports. Riccardi noted the campaign will “spend more money on pizza for volunteers” than on AI tools.
The Deepfake Problem
Thirty-one states have enacted laws regulating political deepfakes. The FEC has not implemented any AI rulemaking. That gap is significant. Without one, campaigns are left to guess what is legal.
Brian Shortsleeve ran for governor of Massachusetts. He used an AI-generated radio ad mimicking Gov. Maura Healey’s voice. The ad said, “We have one of the highest electricity rates in the nation…Thanks to me slapping on excessive fees to fund my climate agenda.”
The ad boosted Shortsleeve’s bid to win the GOP gubernatorial nomination. He lost the September primary in a landslide to Mike Minogue. The ad worked in the short term but could not carry him past the primary.
Spencer Pratt ran for mayor of Los Angeles. He used AI-generated videos portraying himself as a superhero saving dystopian Los Angeles while depicting officials as villains. They grabbed millions of views on X, TikTok, and Instagram.
Pratt became one of the three top contenders for mayor. He was eliminated as he came in third in the June 2 primary. The videos were viral. They did not win.
Why Viral Does Not Mean Victorious
The pattern is consistent. Campaigns spend heavily on AI to create content that spreads quickly. Some of it lands. Most of it does not.
“Utilize[s] every legitimate tool that helps our team improve productivity.”
That is a smart approach. It is also a limited one. AI can produce content faster than humans, but it cannot replace judgment.
The Risks Ahead
The FEC has not implemented any AI rulemaking. That means campaigns are operating without a national standard. Without one, campaigns are left to guess what is legal.
The risk is not just legal. It is reputational. A campaign that produces a deepfake that goes viral will face backlash. The backlash will be louder if the FEC has not told campaigns what is allowed.
The Shortsleeve ad boosted his bid. The Pratt videos got millions of views. Neither candidate won. The pattern is consistent. Viral does not mean victorious.
The Numbers Tell the Story
| Ad | Type | Outcome |
|---|---|---|
| Brian Shortsleeve radio ad | AI-generated mimicry | Boosted bid, lost primary |
| Spencer Pratt videos | AI-generated superhero scenes | Millions of views, eliminated |
| Wesleyan Media Project count | AI-influenced ads | At least 164 this cycle |
More ads, more influence, and fewer wins.
The advice from Ritter is worth repeating: “We have professionals who know how to use and integrate them…to communicate a message, not to show off cool visual tech.” That is the right approach. AI is a tool, not a magic wand.
The pattern is consistent. Viral does not mean victorious.
Where the paper stands
The paper backs disclosure requirements for AI use in political advertising and is against letting campaigns hide their reliance on the technology. Thirty-one states have enacted laws regulating political deepfakes, while the FEC has not implemented any AI rulemaking. The paper opposes broad new rulebooks and rules the biggest firms helped write, which is why it supports narrow disclosure rather than a federal licensing regime that would freeze today’s leaders in place and lock out whoever would challenge them.
Disclosure is voluntary, and campaigns that do not admit using AI are still using it. The Wesleyan Media Project found at least 164 AI-influenced ads this cycle, counting ads that did not disclose AI use. The difference between disclosed and undisclosed AI is real, and the paper wants campaigns to admit what they are doing.
What the reader should watch for is whether disclosure becomes mandatory or whether campaigns are left guessing what is legal. The paper wants the former and opposes the latter. The pattern is consistent: more ads, more influence, and fewer wins.
Source material: “AI is supercharging political ad wars. But viral doesn’t mean victorious,” Fox News.
Get the Notebook.
The day's best stories and every fresh verdict, in plain English, in your inbox by seven. One email a day, no more.