AI is creating a problem where we can’t always trust what happens in real life: voters are doubting everything—whether something is clearly made up (like a funny photo of Joe Biden playing golf) or may actually be true. Just because “you see” or “hear” a piece of content doesn’t necessarily mean it’s accurate.
The New AI Problem for Politics
With generative AI, you can easily create realistic videos, photos, and audio of politicians saying or doing things they’ve never said or done. And some examples of this are already showing up in campaigns and robocalls. Sometimes, those deepfakes will attack an opponent or try to discourage people from turning out to vote. Because these deepfakes move so fast on social media platforms, fact checkers don’t have a chance to catch them before they spread like wildfire.
On top of all this, there is now more awareness of AI among the general public. That has created a greater reason for politicians to say real videos of their opponents were faked. As a result, the public is having trouble agreeing on the most basic facts.
There are now two problems here: false content that looks like it could be real, and legitimate content that can be reasonably disputed.
Regulations Are Behind
While states are trying different approaches to dealing with political deepfakes, the current state of affairs is spotty and uneven. Right now, roughly two dozen states have some sort of law related to AI or deepfakes in terms of elections. Most of those laws restrict using political deepfakes during the last few weeks before an election in order to deceive voters. Others require candidates or campaigns to label their content if it was generated using AI.
Minnesota and Texas have banned sharing political deepfakes in the days leading into an election. Other states are focusing more on labeling requirements or criminal impersonation. Already, courts are challenging how far state laws can go under the First Amendment. For example, judges are raising questions about whether the scope of statutes goes too far, limits free speech, or stifles satire and parody. As a result, many potentially damaging uses that rely on speech remain in a gray area.
The Implications for Voters
Voters are now facing a much larger “verification burden” than existed in prior elections. Today, before accepting vivid video or audio of a politician as factual, citizens increasingly need to wonder: who published the content and which news organization(s) verified the content? Has there been confirmation from other credible organizations regarding the source and context?
Deepfakes have a psychological advantage over traditional misinformation: they appear believable—even after people realize they are fabricated—which makes them effective for propagating information, targeting persuasions, solidifying partisanship, stimulating conspiracy theories, and increasing skepticism toward institutions. All of this can become more serious when hundreds of millions of people are voting in very important elections across the globe.
Practicing Safe Habits to “See Through” AI
To help mitigate the negative effects of deepfakes, voters should consider developing the following habits:
1 – Use reputable news organizations that independently confirm the authenticity of visual and auditory content before spreading it.
2 – Verify shocking video or audio clips via multiple sources, including reputable wire services or prominent local media outlets instead of responding solely based on one viral posting. If something seems to extraordinary to be true, it probably is—especially if it’s not on Google.
3 – When campaign materials utilize AI-created stylization, look for a disclosure label and/or additional context. When you come across sensational material without a disclosure label, treat it as questionable until you receive further information.
4 – Recognize your emotions when viewing deepfakes; their purpose is to elicit strong feelings (such as anger and fear) that bypass rational thought processes.
Opinion: This should be a major concern
Folks, especially on the right, often used deceptive information in the past to sway the masses. I first saw this in 2008 when a video circulated via email claiming Democratic candidates Barack Obama and Hillary Clinton refused to honor the Pledge of Allegiance. I was so angry and confused as to why the mainstream media hadn’t reported such acts of patriotic betrayal. I subsequently searched Google and found out the video had been massively altered to deceive viewers.
After that experience, I investigated almost everything that wasn’t being reported by mainstream news that seemed to be too extraordinary for them to miss. And most of it was fake news.
If deceptive news and information was rampant before the dawn of AI, how big will it be today and tomorrow?
