In this first episode of the CoreyCast, Corey delves into the unsettling world of AI-generated propaganda. As AI becomes increasingly common and integrated into daily life, it is also being exploited to deceive people. Corey discusses a specific instance where an AI-generated video depicting fictitious photos of Hurricane Helene that were convincingly passed off as reality, highlighting the difficulty in distinguishing AI content from real events and the growing potential for AI to mislead the public. Corey also conducts an insightful interview with Mikayla, a fellow VOX Investigates team member, to further explore the implications of AI in the dissemination of false information.
Music Credits:
Exposing The Scam By Adigold
Curious Thoughts By tunestogo
Royalty-free, sourced via Envato
Transcript:
(Edited for Clarity)
Corey: Hello, everyone, and welcome to the CoreyCast, a podcast made by me, Corey. Today’s episode is on how AI is being used to create propaganda.
AI is very common in the world today. It’s convenient, easy to access, and there are little regulations around it. It’s growing and being included in more and more aspects of daily life, but it’s also being exploited.
In the context of today’s podcast, it is being used to trick people. A lot of people might think, “Oh, I won’t get tricked by AI. I’ll spot it. I think it’s easy to spot AI,” but it is not that easy. AI is getting more and more convincing as it’s growing, and it’s only going to get better.
There is one AI video of Hurricane Helene reaching Category 6 and destroying a coastline. People commented on this video, saying it is an entirely real incident in Florida in the US, and 100 times scarier than a tsunami. People like them believe that this video was real. However, there is no such thing as a Category 6 hurricane. The hurricane scale only goes up to a five.
Normally, mistakes like this are what give AI away. Another AI-generated image of Hurricane Helene. Men in the picture had six fingers on one of his hands. AI is not very good with hands, and that is one of the main points to look out for when trying to spot AI.
Corey: So, could you introduce yourself to the podcast?
Mikayla: Hi, I’m Mikayla Johnson. I’m 17. I use she/her pronouns.
Corey: What is your opinion about AI in general?
Mikayla: I feel like it could be really useful and helpful, but I think that it being open to the public, and anybody can use it at any time ,and just have unlimited access to it … I don’t think that’s good.
I don’t think that you should be able to use it like 600 times a day. Maybe everybody could use it, but there’s a limit to how much you can use it.
Corey: Well, do you use generative AI in your daily life?
Mikayla: No, but I have used AI to help come up with titles or just silly stuff. I know there’s a trend going around where you can make yourself into an AI doll. But that’s pretty much it.
Corey: Well, do you have any experience with AI-generated propaganda?
Mikayla: Yes. Thinking back to this past election, the amount of AI videos and clips and things I saw and people believed were insane.
I remember being on Twitter (or X or whatever it’s called) and seeing an AI photo of Kamala Harris, and she was doing like the Hitler salute, and people believed it. It was insane.
I’m like, “How do y’all believe this?”
Corey: Has it ever fooled you before?
Mikayla: Has AI ever fooled me? Yes, but it wasn’t anything serious. I remember it being like a video of a penguin, and I thought it was real, but it was a fake penguin.
Corey: So, what strategies do you use when you think an image is AI and you want to fact-check it?
Mikayla: When I want to fact-check AI, which I have to do all the time for family members, if it’s a picture, hands are so obvious. When you look at AI, it’s not good with hands, feet, or anything. Sometimes, there would be six fingers.
Also, if it just looks too good to be true. I remember Billie Eilish posted a photo when all the wildfires in California were happening earlier this year. It was a little girl in the fire, and the photo looked too perfect. There’s no way there are photographers out there capturing those perfect moments. And she ended up deleting it because people were like, “That’s AI.” And then even celebrities are like, “Oh.” It’s kind of scary, honestly.
Corey: Spotting AI can be difficult, but it’s easy if you know the proper strategies. Here are five ways to spot AI images.
One, look out for all details, like wrong lighting or shadows, and misplaced or weird-looking body parts and words. Also, look out especially for hands, because it is one of AI’s weakest points.
Two, look out for unnaturally glossy or perfect skin. A lot of times, AI will give humans unrealistically perfect skin and hair that just wouldn’t exist in the real world.
Three, look out for good images, but with a blurred background. A lot of times AI just can’t compute the detail of a real photo, so they just blur the background. AI will mainly focus on just the subject.
All three of these imperfections were shown in an AI image of a girl on a boat after surviving Hurricane Helene. Everything other than the girl, the boat she was in, and the dog that she was holding was blurred. The dog had an unnatural leg, and the girl’s face was unnaturally glossy. The image sparked a big debate online with multiple people, including Senator Mike Lee, representing Utah, reposting the image, thinking it was real.
Number four, look for AI watermarks. Sometimes, an AI will put the logo of the company it represents on a photo.
And number five, search the image on the internet. If the image doesn’t appear from multiple sources, or there aren’t similar images with the same person or place, it could be AI.
For years, people have been warning about not trusting everything you see on the internet. Now, it is an even bigger concern because of AI. When you spell out something you think is AI, make sure to carefully check the image so you don’t get fooled.
One Response
Thank you Corey for this insightful and informative podcast. Although I am vaguely aware of AI, I was not aware of the various signs of detecting AI when viewing images.
Continue to educate us as so many decisions are made based upon what we see. Congratulations and Good Job!