In Focus: Experts on growing AI threats

Oklahoma Democratic Party Chair and state lawmaker John Waldron was targeted by an AI-generated audio clip that falsely made it sound like he was making racist comments.

He says forensic analysts later confirmed the clip that made it sound like he was making racist comments toward former Oklahoma Democratic Party Chair Alicia Andrews was fake.

Fake AI audio targets Oklahoma lawmaker

"That was an interesting moment. The audio tape was pretty crudely made and didn't sound like me, and moreover, the words did not reflect who I am, so it's really concerning that somebody was out there representing that as my words and my message," said Waldron.

Waldron says the incident highlights a much deeper issue with the misuse of artificial intelligence in political messaging.

"So it becomes really dangerous when people use a false racial message in politics. It pokes at a wound that we've always had," he said.

Waldron says he plans to introduce legislation at the state Capitol to deal with this issue.

It's becoming harder to tell what's real and what's not because AI is getting better every day. Experts say that people who make this content have a responsibility to use it wisely, but the public also has a responsibility to carefully vet anything they see online.

A former student steps in to help

When Waldron heard the fake, AI-generated audio released of him, he knew exactly who could help.

"The fact that I had this relationship with a former student who knew the history and the context and had the technical skills to address the deepfake really resonates with me. I think it's—it's an example of the universe coming back to me, the energy I put out coming to my defense," he said.

That former student is Tyrance Billingsley. He is the founder of Black Tech Street, a website that covers technology.

He says when he heard the audio, he knew something was off.

"I heard it—didn't even really sound like him. But then, just knowing what I know about AI, having been in this space, I could tell. So it was important to get it kind of run through a couple of different tools. I recommended tools to the Black Wall Street Times and others," said Billingsley.

AI is improving, but so are detection tools

Experts say all it takes is seconds of your real voice to be recorded. Then people can plug it into a website, and before you know it, they've created a fake audio of your voice saying something you didn't even say.

Although it's not perfect yet, Mahdi Khodayar with the University of Tulsa says AI technology is getting better by the day.

"Now it's time to build AI detection models to detect if something is fake or not," said Khodayar.

He says there are things people can look out for when they think something might be AI.

"You can see some odd patterns in the audio signals that can raise some red flag, and you can detect a fake audio. So it can be some odd pauses or some unnatural stops in the audio, or it can be some robotic tone," he said.

The good news is, as AI becomes more advanced, experts say AI detection will become more advanced too.