A video on TikTok appeared to show YouTube influencer MrBeast offering 10,000 iPhone 15 Pro phones to followers for $2 each.
Don’t bother getting your credit card out, because the video was an AI-generated deepfake that somehow evaded TikTok’s content moderation.
MrBeast, whose real name is Jimmy Donaldson, dominates YouTube with 189 million subscribers to his channel. His videos are pretty outrageous and see him doing things like giving away an $800k island or $1m that users need to spend in a minute.
So you’d be forgiven for briefly believing that he’d do something crazy like give away 10,000 iPhones. The $2 upfront payment should have been a hint that this was a scam though.
The video was a little glitchy but it looks and sounds just like MrBeast so it probably fooled plenty of people before TikTok eventually took it down.
Lots of people are getting this deepfake scam ad of me… are social media platforms ready to handle the rise of AI deepfakes? This is a serious problem pic.twitter.com/llkhxswQSw
— MrBeast (@MrBeast) October 3, 2023
TikTok has a verification process in place for new account signups and content moderation policies but none of those caught this fake ad. One comment on MrBeast’s tweet said “Getting verified on TikTok is as easy as taking a nap.”
MrBeast asked “Are social media platforms ready to handle the rise of AI deepfakes?” and the obvious answer is, no they’re not.
In just the last week we’ve seen AI deepfakes of Tom Hanks in a dental ad and Gayle King appearing to endorse a weight loss product.
View this post on Instagram
Getting scammed for $2 or buying a weight loss product that doesn’t really work is the on low end of the potential damage AI deepfakes could cause.
The potential for disrupting elections or spreading harmful disinformation is amplified when the person talking to us appears to be someone that we trust. Companies are trying to develop a digital watermark that identifies AI-generated content but none of them seem to work reliably yet.
Even if they do develop a way to mark AI content, how would they compel people to use it? Until we find a way to instantly know when something is AI-generated we’ll need to be extra cautious about what we believe.
If it’s too good to be true, it probably is.
