LYNCHBURG, Va. (WSET) — Rapidly evolving technology is making it possible to put words in anyone’s mouth.
Using a headshot and AI, our team took ABC 13’s Danner Evans from inside the studio to out in the field.
It takes a matter of minutes to create a video like this, and deepfake videos are popping up all over on your social media scrolls. Most are created by just using an image.
If you think you can tell the difference, you’re underestimating AI.
We snapped a few pictures around town and then had AI generate a deepfake of those photos. We are down at the community market in Lynchburg to see if folks can spot the fake.
Can you spot the deepfake? (Photo: Danner Evans)
Some people could spot the fake of Monument Terrace in Lynchburg.
“Oh, I would say the first one is fake,” Christian Gervais guessed. “Yeah, the top one is fake.”
Others were fooled, like Kayla Gilson, as she sized up the real picture of downtown Lynchburg and the deepfake.
“Oh my goodness,” Gilson said. “I’m going to say that’s the real one.”
Gilson got both of the pictures wrong.
“That’s crazy,” Gilson said.
Can you spot the deepfake? (Photo: Danner Evans)
Dr. Bimal Viswanath is an artificial intelligence researcher at Virginia Tech.
“If you think about this, everyone has the very powerful tool in their pocket,” Viswanath said.
His primary mission is to ensure the safety and trustworthiness of emerging AI technologies.
Viswanath explained that anyone with a smartphone or computer can generate a convincing deepfake in minutes using sites like gemini.google.com.
In a matter of minutes, we uploaded a picture and created a deepfake with his guidance.
Deepfake video of ABC 13 News anchor Danner Evans (Sinclair Broadcast Group).
“This was easy, right?” Viswanath said.
That’s exactly why you may want to think twice about what you post online. Bad players are out there and can easily manipulate your images.
“If you’re uploading your personal images publicly on the web, they can be misused,” Viswanath said. “I would say everyone is at risk at this point.”
Katie Jennings with Lynchburg Police said a bad player could create new content to use in any number of ways to hurt you, like creating fake social media accounts, or making a video that appears you said or did something outrageous that you never did.
“They could create a video for sextortion or blackmail. That would make you feel uncomfortable if you share it, ” Jennings said. “They’re a lot like traditional scammers where they use your emotions against you and try to create a sense of urgency to get you to give them what they want .”
There’s software you can use to protect your photos and videos. If someone tries to alter them using AI, they appear scrambled. However, Viswanath’s team at Virginia Tech found a way around that protection.
Software can protect your images by scrambling the photo if someone tries to use AI to alter it (Photo: Virginia Tech)
“This again becomes a cat-and-mouse game,” Viswanath said.
As we move into the even the next five years with AI, Viswanath has major concerns about deepfakes.
“One thing I’m concerned about is video,” Viswanath said. “They’re getting better and better.”
Forcing all of us to take a closer look and think twice before we believe what we see or make that post on social media.
Think you can spot a deep fake? Take the quiz here.
