Imagine opening your phone and seeing a video of a candidate saying something outrageous. The voice sounds exactly like them. The face moves naturally. The background looks real. You feel your heart rate spike, your fingers hover over the share button, and then you pause—because in the 2024 election season, that video might be a complete fabrication. This is the new reality of political campaigns, and it is exactly why New York Attorney General Letitia James is sounding the alarm. With the midterm elections just weeks away, WAMC’s Grant Ashley sat down with James to talk about how voters can protect themselves from AI-generated misinformation. James, a Democrat, is running for reelection this year against Republican Saritha Komatireddy, a former federal prosecutor. Early voting in New York begins October 24, and Election Day is November 3. But the conversation wasn’t just about campaign schedules or legal strategy. It was about something more unsettling: the way artificial intelligence has made it possible to create believable fake audio, video, and images that can trick even careful voters. The good news, according to James, is that there are still ways to spot the fakes and, more importantly, ways to make sure you aren’t fooled by them. The bad news is that the technology has gotten so good that the old warning signs—strange lighting, unnatural movements, odd audio glitches—are no longer reliable. The fakes have gotten really, really good, James said, and that means voters need to change the way they think about political information.
That’s why James’s office released an updated voter protection guide. The first version, published in 2024, told people to keep an eye out for visual tells like weird shadows or jerky movements. But this year, James and her team recognized that those tips aren’t enough. AI can now make fake videos that look and sound like real people, produce photos that never happened, and generate audio of a trusted politician saying things they would never say. Even worse, AI can create fake messages that appear to come from your local board of elections, complete with official-looking logos and language. And there is another layer to the problem: more and more Americans are turning to AI chatbots for basic questions, including where and how to vote. Chatbots, James warned, can confidently give the wrong answer. They can tell a voter the wrong deadline, the wrong rules for their city or state, or make up voting requirements that don’t exist. The result is confusion, missed ballots, and a quiet but real erosion of trust in the electoral process. So how can voters protect themselves? James said the best approach is old-fashioned: look at the source. If you see information about voting, ask yourself whether it seems reliable. Is it being reported by established news outlets? Is it on your local board of elections website? If a message makes you feel shocked, emotional, or urges you to act fast, be suspicious. Fakes are designed to get exactly that reaction. They want you to panic, to act before you think, or to stay home entirely. The safest move, James said, is to call your local board of elections, call your local elected official, or call the Office of the Attorney General’s election protection hotline at (866) 390-2992. There is no substitute for checking with an official source.
But who is doing this, and why? James was clear about the motive: vote suppression. The goal is to make sure certain people don’t vote, or can’t vote, or are too confused to vote. In 2024, an AI-generated President Joe Biden called New Hampshire Democrats and told them not to vote in the primary. It sounded like him, but it wasn’t. In another case that James’s office was directly involved in, a bad actor surgically targeted Black voters in Upstate New York, Michigan, and Illinois, feeding them inaccurate information about where and when to vote. Michigan Attorney General Dana Nessel helped prosecute the individuals responsible. These weren’t broad, clumsy attempts to spread fake news; they were precise, data-driven efforts designed to exploit specific communities and keep them from the ballot box. This is what makes AI misinformation so dangerous. It doesn’t have to convince everyone. It just has to persuade enough targeted people to stay home or show up at the wrong location. Imagine waiting in line at a polling place only to be told you’re in the wrong district, or receiving a text message claiming that your poll hours have changed. For one person, it’s a frustrating day. For a campaign, multiplied across thousands of people, it’s a strategy. James said the threat isn’t hypothetical or futuristic. It is happening right now, in real time, and her office is working with law enforcement across the country to stop it. The bad actors may be political operatives, foreign agents, or just individuals looking to cause chaos, but whatever their identity, their motives are often the same: suppress the vote, undermine trust, and make people doubt whether their vote matters.
There are laws designed to hold these bad actors accountable. New York State law now requires that political messaging generated or manipulated by AI be clearly labeled. The idea is simple: if voters know a video or message was made by AI, they can take it with a grain of salt. But legal rules are only as strong as their enforcement. James explained that district attorneys are responsible for prosecuting these cases, and the consequences can range from fines to probation to jail time. The law passed recently, and already her office is seeing more fake official notices and manipulated election videos, images, and audio that lack the required label. A lot of bad actors are ignoring the law. Part of the problem is that it can be difficult to figure out where these messages are coming from. AI tools are cheap, accessible, and easily hidden behind anonymous accounts or foreign servers. By the time an election official or reporter catches a fake video, it may have already been shared millions of times. And even when investigators trace it back to a person, proving intent can be tricky. Did the creator make a mistake or deliberately try to deceive voters? For all these reasons, James said, the focus has to be on prevention and public education. Voters need to know what to look for, where to call, and how to report suspicious content. The law is important, but it’s not a shield. In a way, the best defense against AI misinformation is an informed and skeptical public. If people automatically question a viral video that shows a candidate saying something terrible, instead of immediately accepting it, the power of deepfakes begins to fade.
Before Grant Ashley let the attorney general go, he asked a question that must be on many voters’ minds: have fake AI videos come up in James’s own campaign? “Not as of yet,” James said. That’s a reassuring answer, but it also feels like holding your breath. With the election just days away, anything could happen. James herself occupies an unusual position in this story. She is both the official responsible for protecting voters from AI misinformation and a candidate on the ballot, asking those same voters to trust her. She has to navigate the same digital landscape as everyone else, and she knows that a single well-timed deepfake could go viral and dominate the news cycle in the final days of the race. So far, she has been lucky. But luck isn’t a strategy. Her office’s guide, hotline, and public outreach are all part of a larger effort to make sure New Yorkers are ready, not just for this election but for every election to come. When Ashley thanked her for making time, James responded with a reminder that cut through all the technical complexity: “I appreciate you, and make sure you go out and vote.” It’s a simple message, but in an age of AI-generated robocalls, manipulated videos, and confident chatbots, it’s also a radical one. Voting is the moment when each person independently chooses to participate in democracy—not through a screen, not through a shared video, but by showing up in person or by mail, with a ballot and a voice.
In the end, the interview with Letitia James captured a central challenge of modern elections: the problem isn’t just getting people to vote; it’s making sure they can vote with accurate information. AI has made it cheaper, easier, and faster to create misinformation than ever before. A convincing fake can be produced in minutes and spread to millions within hours. But the defenses haven’t changed. Check the source. Call the local board of elections. Don’t trust a shocking video just because it looks real. If you see something suspicious, report it to the Office of the Attorney General’s election protection hotline at (866) 390-2992. And above all, don’t let anyone—human or artificial—talk you out of casting your ballot. The technology may be new, but the values are old. Every generation has had to deal with rumors, lies, and manipulation at the ballot box. What makes this moment different is that the lies can now wear the face and voice of someone you trust. That’s scary, but it’s not unbeatable. James’s final message was simple: go out and vote. Do your part. Verify before you share. And remember that the most human act of all is the quiet, deliberate decision to show up on Election Day and make your voice heard. A deepfake might look and sound like a politician, but it will never be able to cast a ballot. Only you can do that.