Peggy Flanagan calls out AI deepfake used against her in political ad
Peggy Flanagan was not expecting to see her own face being used against her in a political ad — and definitely not like this.
The lieutenant governor of Minnesota and candidate for the United States Senate went public to denounce an attack that blends politics with technology in a way many people still think of as science fiction: an AI deepfake using her likeness, created to mislead voters during the Democratic-Farmer-Labor Party primary, known as the DFL. 😮
The story blew up immediately — and it is not hard to see why. When a candidate for federal office has to warn voters that the face they are about to see on TV is not actually hers, something has shifted in American politics. And Minnesota, which already has a specific law regulating deepfakes in elections, became the perfect stage to test just how far that legislation can go when the technology actually shows up for real.
What happened to Peggy Flanagan
The episode involving Peggy Flanagan started gaining traction when the candidate herself posted a direct message on social media, alerting voters that they might see a TV ad featuring someone who looks like her but is not actually her. Flanagan was clear in stating that her opponent’s Super PAC was using an AI deepfake to deceive Minnesota voters, and that the strategy exists because the other side cannot win on the truth.
The ad in question was produced by North Star Dawn, a PAC supporting congresswoman Angie Craig, Flanagan’s rival in the race for the DFL nomination. It is important to note that North Star Dawn has no direct ties to Craig or any other candidate. A spokesperson for the congresswoman stated that Craig does not support the use of artificial intelligence in political ads.
On the other side, Jerid Kurtz, a spokesperson for the PAC, called Flanagan’s accusations baseless. In an official statement to WCCO, Kurtz said the group will continue to follow all federal and state laws while informing Minnesota voters about the millions in corporate dollars that special interest groups are allegedly spending to elect Flanagan.
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
What makes this case particularly striking is the level of sophistication of the technology involved. We are not talking about a sloppy edit that anyone can spot in two seconds. The generative AI models available today can produce videos and images with startling realism, syncing lip movements, facial expressions, and vocal tone so convincingly that the final product looks absolutely legitimate to anyone who is not actively looking for signs of manipulation. For an average voter watching an ad during a commercial break, the odds of noticing it is fake are extremely low — and that is exactly where the real danger of this technology lies when it is applied to politics.
Flanagan did not just call out the incident — she also used the exposure to emphasize how important it is for Minnesota voters to stay vigilant about the content they consume during election season. The candidate made it clear that this kind of attack will not intimidate her into dropping out of the Senate race, but she publicly acknowledged that the situation creates a new and serious challenge for anyone running for public office in the age of artificial intelligence — because now it is not enough to defend your ideas, you also have to defend your own image against fake versions of yourself. 😤
Minnesota already has a law against deepfakes in elections
Minnesota did not arrive at this moment unprepared. The state passed specific legislation during the 2023-2024 legislative session regulating the use of AI-generated deepfakes in the electoral context. The law was the result of a bipartisan effort, bringing together legislators from across the political spectrum around a shared concern: preventing technology from being used to manipulate election outcomes.
According to the text of the law, a deepfake is considered illegal when it is so realistic that a reasonable person would believe it depicts speech or conduct by an individual who, in reality, never made those statements or behaved in that manner. In addition, for there to be a violation, the material must meet several specific criteria:
- It must be produced without the consent of the person depicted
- It must have been created with the intent to harm a candidate or influence the outcome of an election
- It must be distributed within a period of 90 days before a party convention, or after the start of the early voting period before presidential primaries, state or local primaries, or regular or special general elections
The Peggy Flanagan case puts this legislation directly to the test, exposing the practical limits of any law when it faces technology that evolves faster than legislative processes can keep up. The central question is: does the ad produced by North Star Dawn meet the criteria defined by the law? And if it does, what mechanisms are available for a response fast enough to prevent irreversible damage to the candidate’s campaign?
The biggest challenge is not just identifying who produced the content — which is already a complex task given the anonymity that digital tools allow — but also acting quickly enough to contain the damage before the material spreads beyond control. In elections, time is a critical factor. An AI deepfake released in the days leading up to a vote can circulate for hours or days before it is officially taken down, and in that window it will have already reached a significant number of people. The question that lawmakers in Minnesota — and across the country — need to answer is: how do you create rapid response mechanisms that can keep pace with the speed at which disinformation spreads across digital networks?
Experts in digital law and election technology point out that laws like Minnesota’s are a critical step forward, but they need to be backed by technical infrastructure and dedicated human resources to monitor and act on violations in real time. It is not enough to ban something on paper if there is no operational capacity to enforce and punish effectively. The episode with Flanagan serves as a real-world case study both for states that have not yet regulated the issue and for those that already have but have not yet faced a concrete situation requiring direct enforcement during an active election cycle.
What experts are saying about deepfakes in 2026
Dr. Manjeet Rege, a professor and director of the Center for Applied Artificial Intelligence at the University of St. Thomas, brought an important technical perspective to the conversation. According to him, deepfakes are becoming increasingly difficult to detect — and that applies even to people who work with technology every day.
In a statement to WCCO, Rege was direct: AI-generated content is becoming so realistic that even experts need time to tell what is real from what is fake. That statement is particularly alarming when you think about the electoral context, where most voters do not have a technical background and consume political content quickly, without the time or tools to do a detailed analysis.
Rege shared some practical tips for anyone trying to spot deepfakes in everyday life:
- Watch the edges of the face and the hairline — these are the areas where AI tends to leave the most visible artifacts, such as overly smooth transitions or subtle distortions
- Pay attention to background elements — objects or settings that look slightly off from what you would expect can be an indicator of digital manipulation
These are simple tips, but they can make a difference when evaluating the authenticity of a political video. The problem is that as the technology advances, even these visual identification techniques will lose effectiveness, requiring automated detection tools to keep up with the pace of generative model evolution. 🔍
The race between deepfake creation and detection
There is a curious and concerning dynamic playing out in the artificial intelligence space: the same advances that make deepfakes more convincing are also fueling the development of more sophisticated detection tools. It is a kind of technological arms race, where each side tries to outpace the other. Tech companies and academic research centers are investing in algorithms that analyze patterns invisible to the human eye — such as inconsistencies in pixel compression, anomalies in lighting patterns, and imperceptible variations in video frame rates.
However, as Dr. Rege himself pointed out, these tools are still not foolproof. And in the context of an election campaign, where content circulates at breakneck speed, detection needs to be nearly instantaneous to have any practical effect. The gap between a deepfake being published and being identified as fake can be enough to shift thousands of voters’ perception of a candidate.
Why this matters far beyond Minnesota
What happened to Peggy Flanagan in Minnesota is not an isolated incident — it is a clear signal of what is coming for politics on a global scale. As AI deepfake tools become more accessible, cheaper, and easier to use, the barrier to entry for creating sophisticated electoral disinformation drops dramatically. Before, producing a convincing fake video required specialized technical teams, time, and money. Today, generative AI platforms allow anyone with internet access and some basic know-how to produce manipulated content in a matter of minutes. That completely changes the risk landscape for candidates, parties, and most importantly, for voters who depend on available information to make informed decisions at the ballot box.
In the United States, the debate over AI regulation in elections has gained momentum over the past two years, but legislative progress remains fragmented. Some states, like Minnesota, got out ahead with specific legislation. Others still rely on generic defamation laws or the terms of service of digital platforms themselves to try to contain the problem — which in practice means much weaker protection. The absence of clear federal regulation creates a patchwork of rules that varies from state to state, further complicating any coordinated response when a case like Flanagan’s emerges during an actual electoral race. 🗳️
The voter’s role in the age of AI
For voters, this landscape demands a new kind of digital literacy — the ability to question the authenticity of the political content they consume before forming their opinions. It is not about distrusting everything, but about building the habit of checking sources, seeking confirmation through official campaign channels, and watching for technical signs that might indicate manipulation, such as lip movements slightly out of sync, inconsistent lighting, or artificially uniform image quality.
It is a new responsibility that falls on voters — one nobody asked for, but one that came along with the age of artificial intelligence. The Peggy Flanagan case makes that crystal clear: the technology has already arrived at the ballot box, and ignoring that fact is no longer an option.
Meanwhile, the DFL nomination battle between Flanagan and Craig continues to be one of the most closely watched races in American politics in 2026. Not just because of the clash between two prominent party figures, but because the deepfake episode has turned the race into a real-world laboratory for the ethical and legal limits of using artificial intelligence in democracy. What happens in Minnesota could end up setting precedents for the rest of the country — and maybe even beyond. 🤖
