Defending Democracy: The Case for Regulating Political Deepfakes in the United States
Modern AI makes deception easier than ever. It's time to act.
In 2024, voters in New Hampshire got a phone call that sounded like a presidential candidate telling them not to vote in the primary. It wasn’t real. But it was believable, and that’s the problem—because if you believed it, your vote was gone.
Deception in politics isn’t new. But for the first time in history, artificial intelligence makes deception easy, scalable, and nearly impossible to detect.
With AI, anyone can clone voices, create realistic videos, and spread convincing political messages to millions in minutes. Those messages look real enough to act on. And when voters can’t trust what they hear, they can’t make informed choices.
The truth is simple: The United States needs to act now before deceptive AI in elections spirals out of control.
AI-generated political content is uniquely dangerous. Deepfakes look completely real, and we aren’t prepared to recognize them. Even experts struggle to identify deepfakes, and nearly 40% of people over 65 didn’t know what deepfakes were. We’re used to trusting what we can see and hear—but that assumption no longer holds.
And thanks to social media, a deepfake can reach millions before anyone has time to prove it’s false.
But the real danger isn’t that one fake video fools someone—it’s what happens after. People will begin to doubt everything—even what’s real. If any video could be fake, how can someone know what isn’t? When voters can’t agree on reality—democracy breaks down.
In 2024, an AI-generated video made Kamala Harris appear to say offensive, controversial things. It reached over 100 million views before it was labeled as a parody. But by then, the damage was already done. When something looks and sounds authentic, people treat it as real.
These videos often appear exactly when they can do the most damage, like right before an election.
Right now, we aren’t ready. Not as voters, and not as a country. Only 28 states have laws that regulate political deepfakes, and 26 of those states only require labels on AI-generated content. But labels are easy to miss, and even easier to ignore.
California’s main deepfake regulation law was blocked in 2025, leaving our state with little to no regulation. And those other 22 states have nothing, leaving bad actors with fewer barriers. A single convincing deepfake released at the right time could create widespread confusion—not because everyone believes it, but because no one knows what to believe. The goal may not be to change votes directly, but to make Americans doubt the results entirely.
When people no longer trust election outcomes, democracy begins to fall apart. AI deepfakes are already blurring reality in the 2026 midterms. In one popular video from March, a Texas State Representative appears to say, “Radicalized white men are the greatest domestic terrorist threat in our country.” The representative never filmed that video—it was generated using AI.
We need a clear national standard. AI-generated political content that misleads voters—whether by giving false voting information or impersonating a candidate—should not be allowed in political advertising. Social media platforms should be required to remove it. At the same time, obvious satire and parody should be explicitly protected.
This is not a radical idea. We already require platforms to remove harmful or illegal content, and we already prohibit forms of fraud and voter deception.
The First Amendment protects speech—it doesn’t protect impersonation. And free speech only matters if we can trust the system that protects it.
Finally, there is something each of us can do. Pause before sharing something shocking. Check the source. And if it’s deceptive—report it. Responsibility falls on us to protect what is real.
Democracy depends on trust: trust in information, in institutions, and in each other. As they continue to improve, deepfakes threaten that trust in a way we have never seen before, and right now, we are not ready to face them. The 2026 midterms and 2028 presidential election are approaching. If we fail to act, the foundation of free elections begins to erode, and we risk a crisis where Americans no longer believe in the outcome of elections. We don’t need a perfect solution—we need a clear one. A national standard. And we need it before it’s too late.
Note: This piece was adapted from a speech I wrote and delivered in early 2026.