Rebecca Cooke isn’t suing anybody. Not yet, anyway. But this week the Democratic candidate running for Congress in western Wisconsin’s 3rd District did the next best thing – she sent her Republican opponent, Rep. Derrick Van Orden, a formal cease and desist letter over a string of AI-generated ads that put words in her mouth and scenes in her life that never happened. Literally never happened. They were made up by a computer.

So What Did Van Orden Actually Do?

Here’s the thing that makes this one different from your garden-variety campaign mudslinging. Van Orden’s team didn’t just stretch the truth about Cooke’s record or clip a soundbite out of context – the oldest trick in the political ad book. They went further. They used AI to generate deepfake content depicting Cooke in situations that are, as far as anyone can tell, entirely fictional. One ad shows her in an invented conversation with former President Joe Biden. Made from whole cloth. Pixels pretending to be reality.

Cooke didn’t mince words about it. She said the deepfakes “undermine our democratic process,” which, yeah – I’d say that’s putting it mildly. We’re not talking about a Photoshop job that makes a politician’s tie look a little crooked. We’re talking about synthetic media designed to make voters believe they’re watching something real when they’re not.

Why This Feels Different

I’ve covered a lot of nasty campaign seasons. Attack ads are basically a genre at this point, with their own tropes – grainy black and white photos, ominous narrators, that one unflattering freeze-frame every campaign seems to dig up. But there’s always been an unspoken floor to how low things go, because at the end of the day the footage was real, even if the editing was dishonest. AI breaks that floor. It doesn’t just distort the truth, it manufactures an entirely new one.

Is This Even Legal?

Good question, and honestly? It depends who you ask. Wisconsin, like a lot of states, has been scrambling to catch up with AI in political advertising, and the legal landscape (sorry, I know I’m not supposed to use that word, but it genuinely fits here) is messy. A cease and desist letter isn’t a lawsuit. It’s a warning shot – a formal “knock this off or we’ll see you in court” that doesn’t carry the force of law on its own. Van Orden can ignore it. He might ignore it. Candidates ignore these things all the time and nothing happens.

“These deepfakes undermine our democratic process.”

But here’s what I think people are missing in all this. Whether or not the letter has legal teeth isn’t really the point. The point is that we’re now at a stage in American politics where a sitting member of Congress is apparently comfortable using fabricated AI video of his opponent to try to win a race. Not altered footage. Not a misleading edit. Fabricated. And that should bother you regardless of which party you vote for, because the next deepfake might not be about Rebecca Cooke. It might be about a candidate you actually like. It's a tactic that's popped up elsewhere too, like Warner Bros.' recent cease and desist letter against Fox Nation over a blocked Chris Harrison show.

The Bigger Mess Nobody’s Fixed Yet

This is the third House race this cycle – at least that I’m aware of – where AI-generated content has become a central campaign flashpoint, and we’re still basically in the Wild West phase of regulation. A handful of states have passed disclosure laws requiring campaigns to label synthetic media. Congress has talked about doing something nationally. Talked. That’s about as far as it’s gotten. On the regulatory side, a federal appeals court just dealt AI companies a blow by closing a major copyright loophole.

Meanwhile the technology keeps getting better and cheaper, which is a genuinely terrifying combination if you think about it for more than five seconds. A few years ago making a convincing fake video of a political opponent required real resources – skilled editors, decent budgets, time. Now? Anybody with a laptop and the right software can do a passable job in an afternoon. The barrier to entry basically collapsed. And when the barrier to entry collapses, the volume of bad behavior goes up. That’s just how it works. That same unsettling trajectory is already playing out with Meta's new AI agent, Muse, which millions downloaded last week.

Van Orden’s camp hasn’t exactly been shy about leaning into combative tactics before – this isn’t a guy known for playing it safe or soft. So I’m not holding my breath that one letter from Cooke’s team is going to make the ads disappear. But I do think this moment matters beyond just this one district in Wisconsin, because it’s forcing a conversation that voters, platforms and lawmakers have all been avoiding: what happens when “seeing is believing” stops being true in politics?

What This Actually Means

Look, I don’t know how this particular standoff ends. Maybe Van Orden pulls the ads quietly and nobody talks about it again until the next controversy. Maybe Cooke actually follows through with litigation, and this becomes one of the first real legal tests of AI deepfakes in a congressional race – which, if I’m being honest, might be exactly what needs to happen for anyone to take this seriously. Sometimes it takes one messy, high-profile case to force the issue, the same way a handful of early internet privacy lawsuits eventually dragged tech companies toward (actual, if imperfect) accountability.

What I do know is that this isn’t going away. Not from this race, and not from American politics generally. The 2026 midterms are shaping up to be the first real test of what happens when campaigns have access to tools that can manufacture reality itself, cheaply and at scale. Rebecca Cooke’s cease and desist letter might end up being a footnote. Or it might be the opening chapter of a much bigger fight nobody’s ready for. Either way, I’d keep an eye on western Wisconsin. This probably isn’t the last we hear about it.