The integration of artificial intelligence into our daily lives has felt like a double-edged sword, bringing newfound convenience while simultaneously ushering in a era of uncertainty regarding digital truth. Nowhere is this tension more palpable than in the volatile arena of Minnesota politics, where a recent campaign advertisement targeting Lieutenant Governor Peggy Flanagan has ignited a fierce debate. As Flanagan seeks the U.S. Senate seat vacated by the shifting tides of the DFL platform, a super PAC supporting her primary rival, Representative Angie Craig, released an ad that has sparked significant backlash. Dozens of DFL legislators have publicly denounced the production, arguing that it relies on AI-generated “deepfake” technology to distort reality and mislead voters. This incident marks a critical juncture where the speed of technological innovation is colliding head-on with our traditional expectations of political integrity, leaving voters and officials alike to scramble for a way forward.
To understand why this digital manipulation is so unsettling, we must consider the sophisticated nature of these tools. Manjeet Rege, a professor of software engineering and an expert in applied AI, has spent time demonstrating exactly how convincing these fabrications can be. By simply capturing thirty seconds of a subject’s speech and mannerisms, AI can synthesize a version of that person speaking languages they do not know or acting in ways they never actually behaved. Rege points to specific “tells”—such as unnatural eye contact, fixed gazes, or subtle inconsistencies in background movement—as the best defense for the average viewer. Yet, even with these clues, the technology is advancing at a breakneck pace, making it progressively more difficult for the public to distinguish between a genuine recording of a candidate and a computer-generated puppet.
The central issue here is whether this specific advertisement crosses the legal boundary established by the Minnesota Legislature just last year. In response to the growing threat of election interference, lawmakers passed legislation aimed at curbing the use of deepfakes in political advertising, specifically targeting content that might deceive a “reasonable person.” However, the problem remains that this law is entirely untested in a court of law. As political analyst Blois Olson notes, we are currently in uncharted territory. While it is easy to condemn the concept of deepfakes, applying the statute requires a precise definition of what constitutes a manipulated reality versus what constitutes traditional, albeit aggressive, political hyperbole. We have always had attack ads, and we have always had exaggerations, but the speed and ease with which AI can manufacture false scenarios have left our current legal framework struggling to catch up.
The political fallout from this ad has created a complex web of blame and deflection. While the super PAC—an entity inherently separate from the official Craig campaign—is the source of the video, the association has forced the Congresswoman’s camp to issue a firm statement clarifying that she does not support the use of AI in political advertising. Meanwhile, Lieutenant Governor Flanagan’s team is actively exploring legal options, attempting to determine whether the state’s nascent anti-deepfake laws provide grounds for a successful challenge. This process is inherently messy; it’s a high-stakes game of cat and mouse where the candidate who is targeted must decide whether to lean into a legal battle that might further circulate the offensive video or to focus on defending their record through conventional campaigning.
Beyond the legalities, there is a profound human element to this story that concerns the sanctity of the democratic process. When voters are bombarded with content that looks and sounds like a candidate but isn’t actually them, it erodes the foundation of trust upon which our elections rely. It creates a “liar’s dividend,” a situation where even authentic footage can be dismissed as fake by bad actors, leading the public to eventually stop believing anything they see at all. This isn’t just about Minnesota or one specific primary race; it’s about the future of political discourse. If we cannot agree on a baseline of shared reality, the ability for a candidate to communicate their vision—or for an opponent to critique it fairly—becomes fundamentally compromised.
Ultimately, we are currently witnessing a “coming-of-age” moment for election regulations. The upcoming August primary will likely serve as a litmus test for how Minnesota handles this new digital landscape. Whether the legislature needs to refine the law to provide clearer definitions, or whether the courts must decide the limits of free speech in an AI-driven age, one thing is clear: the genie is not going back into the bottle. As technology continues to outpace our laws, the responsibility may fall increasingly on voters to remain hyper-vigilant. We are moving toward a future where, more than ever, we must question the provenance of what we see on our screens, lest our democracy be shaped not by the voices of those running for office, but by the algorithms that simulate them.