Tips from U of I experts on how to protect yourself from deepfakes during election season

AI-generated “deepfake” images of political candidates have been used by advocacy groups to spread false messages on social media. But there are ways to protect yourself from misinformation, said Mary Ton, an assistant professor and the digital humanities librarian at the University of Illinois at Urbana-Champaign.

With the election less than a month away, deepfakes are a growing concern. AI-generated images, videos and audio recordings are being used to spread misinformation, at a time when voters are relying on digital media to learn about candidates and their campaigns. 

University of Illinois digital humanities librarian and assistant professor Mary Ton said she shares the concerns of some who’ve spoken up against the publication of manipulated videos.

Taylor Swift came out and made a statement because she was worried about the potential impact of AI-generated images that were advancing a view that was not hers,”  Ton said.

It’s important to stop, investigate, look and listen when you encounter an image or a political view posted online, she said. And it’s a good idea to pause before you repost something that may or may not be true. 

Other advice for evaluating the credibility of an image: pay attention to minor details, check a variety of sources and stay cautious, Ton said.

Arindam Banerjee, who researches AI as a professor at the Grainger College of Engineering, encourages people to notice the lighting and shadows in a photo or video of a room, to help identify if it’s real. 

“If you take a picture outside you will see that the lighting and the length of the shadow and the alignment, there will be a perspective to it,” Banerjee said. “Deepfakes can’t maintain those geometries, because they don’t understand yet there is sun shining.”

Deepfake content can cause lots of harm by spreading false messages, and there are concerns about the impact on journalists and artists, whose creative work is used to generate manipulated content. However, there are some benefits to AI-generated content as well, Ton said.

Deepfakes play a crucial role in the entertainment industry, serving as the basis for most of the visual effects we see today, she said. One example is the early deepfake technology used to age and de-age actors on screen. The same applies to the dubbing process. 

“It creates a more seamless experience for viewers of a movie in other languages in that they can more naturally match the lip movement of the actor to the spoken word,” Ton said. 

With potential for both harm — and good — Ton said it’s important for people, especially content creators, to keep up with technology. 

“Explore AI because this is a technology that is evolving,” she added. “We’ve always used a wide range of tools to help us tell really good stories and express our thoughts about the world and there are lots of exciting opportunities that engaging with this technology can present.”

Illinois Student Newsroom

At the IPM Student Newsroom, journalism students from the U of I's College of Media work alongside professional journalists -- public radio reporters, editors and producers -- to produce multimedia stories on issues affecting east-central Illinois. Follow on Instagram: Illinois Student Newsroom