Deepfake Ads Are Flooding Campaigns – Here’s How to Spot Them

Photo: Shutterstock

AI-made political ads are spilling into our feeds faster than the rules that should label them.

Story Snapshot

States moved to require AI disclosures, but rules vary and face court fights
A viral Los Angeles mayoral video used synthetic imagery with unclear sponsorship
Most laws focus on labels, not bans, and enforcement remains uneven
Analysts see little proof these ads change votes, but they do shape attention

Unlabeled AI Political Ads Are Now Part of the Campaign Toolkit

Campaign-style videos made with artificial intelligence now mix fantasy visuals with real politicians and hit social media at speed. One Los Angeles mayoral video cast Spencer Pratt as a caped hero while turning Mayor Karen Bass into a comic-book villain and Governor Gavin Newsom into royalty. The production company’s founder teased the ad, posted it the next morning, and Pratt shared it. Neither the company nor the campaign explained who paid for it, leaving key sponsorship questions open.

That missing paper trail is the point. When a video looks real enough to share but has unclear funding, voters cannot weigh the source. That undercuts a basic norm of American elections: know who is asking for your vote and why. Disclosure laws exist to protect that norm. Without clear labels, attention can be captured by spectacle, while accountability lags behind the view count. Voters deserve both the show and the credits.

States Are Writing Rules, But The Map Is Patchwork

By late 2025, more than two dozen states moved on political deepfakes. Most picked a simple fix: require a plain label when artificial intelligence alters an ad. Minnesota and Texas drew brighter lines by blocking certain deepfake drops right before elections. The trend shows lawmakers view unlabeled synthetic content as a concrete risk that calls for basic transparency, even if they stop short of bans.

California offered a crisp script: say “Ad generated or substantially altered using artificial intelligence.” That kind of short, direct notice is easy to read and hard to miss. Yet the state also shows the legal friction. Some provisions drew First Amendment and internet liability challenges. Courts paused one measure and struck down another, raising the stakes for how laws are drafted and enforced. The result is a checkerboard of timing windows, definitions, and remedies across states.

Disclosure Versus Deception: Where Common Sense Draws The Line

Campaigns and consultants push back on blanket labels. Some argue that satire is obvious and should not need a warning. Others warn that labels can make honest ads look shady to viewers. Their fix is simple: label only when content would mislead a reasonable person. That stance matches a core conservative value: target the bad act, not the tool. If the problem is deception, write rules that punish deceivers, not creativity.

Voters can take that test to the feed. If a video shows a politician saying words they never said, or in a scene that never happened, it deserves a clear tag. If it is a cartoonish spoof, treat it like a late-night joke, not breaking news. Some groups already pair written and spoken labels to avoid confusion, which sets a workable floor for good-faith actors. That still leaves a gap for third-party hit jobs that hide their hand.

What We Know, What We Don’t, And What To Watch Next

Evidence that AI ads change votes is thin. Analysts say these spots can raise money, fire up supporters, and earn media, but hard proof of shifted ballots is scarce. That does not make them harmless. It means their main power today is attention, not persuasion. A label regime is a proportional first step: let the speech flow, but flag the method so viewers can add context before they share or donate.

Lawmakers should focus on three practical moves. First, require short, standard labels on altered political ads across states to end the patchwork. Second, set pre-election rules that target intent to deceive, not satire, to pass constitutional tests. Third, demand platform logs for political ads with synthetic media so investigators can trace sponsors and reach. That blend protects speech, deters fraud, and gives voters what they need most in a noisy year: the truth about what they are watching.