Tom Hanks says AI version of him used in dental plan ad without his consent

MicroWave@lemmy.world to News@lemmy.world – 650 points –
Tom Hanks says AI version of him used in dental plan ad without his consent
theguardian.com

Tom Hanks has warned fans that an ad for a dental plan that appears to use his image is in fact fake and was created using artificial intelligence.

In a message posted to his 9.5 million Instagram followers, the actor said his image was used without his permission. “BEWARE!! There’s a video out there promoting some dental plan with an AI version of me. I have nothing to do with it,” Hanks wrote over a screenshot of a computer-generated image of himself from the clip.

The Oscar winner has expressed concerns in the past about the use of AI in film and TV, although he has not shied away from approving digitally altered versions of himself in film.

117

You are viewing a single comment

Between voice cloning, gpt-4, and social media, the technology exists TODAY for scammers to call you at 4am with the voice and intimate knowledge of a loved-one, and tell you that they need you to send them money for an emergency.

You thought old people were easy to scam before? We're about to enter a golden age of manipulation.

My grandmother is a Greek immigrant and doesn't speak English very well. Back in 2017 someone called her and told her myself and my mother were dead and she spent like 3 days freaking out and crying. I can only imagine what would happen if someone were to do that today while emulating a voice she knew...

I read something recently that said a study was performed and Teens are most susceptible to getting scammed online.

Its because their brains are still developing. As they get older, they'll gain more common sense (hopefully)

I'm guessing that's because of sheer numbers, there are going to be more teens online than any other demographic.

... Or you just have a word/phrase known between yourself and the other person.

They would fail. I have many ways to corroborate and they know I don't answer the phone and to call 911. Nobody is calling me in a emergency ever. I don't drive, I don't give money to anyone etc... People need to have the policy of if you can make a call and it's an emergency then call 911. Also always have roadside assistance.

Didn't this happen recently? I thought I saw an article about a guy in India who got an ai-faked video call from a friend who needed a quick money transfer to get out of a tough spot. He said it was sketchy but believable. It's already here

And the answer is to get off the internet and start trusting those close to you more than digital strangers.

Yes, the technology exists. No, it's not a threat for your grandma. Scammers would first need need to know which phone number is your grandmas, them they need to find out the relatives of your grandma, obtain enough sample data from your voice and train an AI model for at least a few hours to imitate your voice. That's not a realistic scenario to do for a slim chance of getting a few thousand bucks. This kind of social engineering attack is only viable for very rich persons and businesses.

I'm sorry, but your assessment of how difficult that would be is WAAAAAY off.

Scammers are already doing stuff like this en masse with highly customized email scams.

The way this scam would work is to start with YouTubers, where grabbing the voice data is easy. Then you find their Facebook profile... Very easy, since people use the same usernames, or they go out of their way to link their profiles.

It's a pretty easy step to make friend requests with those people. And then a very easy leap to find their relatives real names and towns through their Facebook connections.

Now you take their connections and towns and do reverse phone number lookups.

ALL of this can be automated. Every step.

The voice cloning and gpt-powered phone calls can be automated now, too.

The only reason this isn't happening at scale is that scammers haven't had enough time to adapt yet.

It's weird you talk about how easy it is but your only example is with very public people where all you need is a Google search to get their info.

I'd guess that most people with public social media accounts would be susceptible to something like this. As long as there are videos available with the person speaking, which are plentiful by way of instagram reels / tiktoks, the rest of what the commenter described above sounds totally feasible.

You don't need another example if you understand how many people are covered in that category.

Do you have any idea how many people have at least 5 minutes of audio on YouTube? (That's all you need for voice cloning), tens of millions? Hundreds of millions?... And how many of them have a Facebook, insta, tic Tok, or Twitter account? Virtually all of them.

If you wrote a script to do what I outlined, it would run FOREVER, because he users would be signing up and making videos faster than this script could ever hope to keep up.

People are reaaally downvoting you, but how would someone call my G-Ma and imitate my voice using AI?! My voice isn’t on the internet. That’s an insane thing to fear for any regular person.