You may have seen recently advertisements on “The Pink Salt Diet”, the latest weight loss gadget with elegant celebrities and doctors.
Lisa Rozner of CBS News New York discovered that some of these advertisements are generated by AI, known as Deepfakes, and bad actors use the resemblance of a Manhattan psychologist, to claim false approval.
Controversy “The Pink Salt Trick”
Dr. Rachel Goldman regularly educates the public on obesity and has reached a summit in her career when she appeared alongside Oprah Winfrey on “Oprah Daily”, starting in 2023. She has since appeared on several episodes.
But this world exhibition also made it the target of Deepfake videos manipulated digitally and generated by AI. She said that she feared that some people to take harmful medical advice, thinking that it came from her.
Goldman, who is also a deputy professor at NYU Grossman School of Medicine, said it was difficult to have the videos removed, so she issues a warning to the public.
In an episode in 2023, Goldman told Winfrey to talk to people with obesity.
“We do not want to define an individual by their state of health,” she said.
CBS News New York
From this same episode, a version generated by Ai-Ai was created in a way and she says seen by her followers on Facebook and YouTube with her saying instead: “The tip of pink salt is 100% natural and free from side effects.”
“It looks like me. It’s a bit to talk to me, but it’s not me at the same time,” said Goldman. “It’s very weird and scary.”
But even more frightening Goldman says people fall in love. She showed CBS News New York a text from a parent who said, in part, “does it work? Send me the recipe.”
Goldman says that at least two people had sent her by e-mail that they had bought the product, including a woman in Australia.
A customer sent her a message that she was billed triple and asked: “Do you have a product or does he have everything?”
“Every day, I receive messages,” said Goldman. “I want them to know, don’t buy that. It is not something real that I approve. That’s wrong.
“I really try to come back to as many people as possible and to spread more knowledge before more people buy it or try something that could be dangerous, right? We just don’t know,” she added. “Supplements are not regulated by the FDA. I tend to say that everything that is a quick solution or a hack, is generally not true.”
Deepfakes Target Oprah Winfrey
Goldman says that the false version of AI even shared false medical references, as it frequented Stanford, when in reality it studied at Farleigh Dickinson University.
She said that she is worried about the dangers of false medical approvals like this.
“Some people even share their medical history with me in an email or in a telephone call, asking if it is sure to them,” said Goldman.
CBS News New York
The task of telling the truth and removing the deep buttocks is insurmountable, even for a media magnate like Winfrey. A representative said that she had a digital security team that works every day on withdrawals from false advertisements and channels.
CBS News New York came across the advertisements generated by AI on YouTube. In one, a Winfrey video is manipulated to make him say: “This morning, I watched the exact place where I keep my pink salt recipe.”
Winfrey wrote on Oprah.com in August: “Let me say clearly: if you see an ad with my face on a” product “, that’s wrong.”
Meta and YouTube told CBS News New York that they had deleted the accounts and advertisements linked to Goldman and Winfrey who violated their policies.
“It was very disturbing for me,” said Gayle King
In 2023, the co-host of CBS Mornings, Gayle King, went to social networks to denounce a video of Deepfake AI, wrongly describing his approving a weight loss product.
CBS News New York
CBS News New York asked King: “How was it for you?”
“Well, think about that, Lisa, I didn’t even know it. People started calling me, asking me, where can I get it?” Said King. “And then I watched it and I thought:” Well, that’s the outfit I wore. ” That is to say, it looks like my voice. It looks like me.
“But if you look carefully, but you have to look at very close to seeing that the lips are slightly extinct and the way I speak is not words that I would use. But it was very disturbing for me,” added King. “I would just hate people buy something because they think I approved it when I have nothing to do with that, and I don’t know it’s the safe thing, you know? What are the ingredients. I didn’t know anything.”
Legislation designed to face Deepfakes
The member of the Congress Yvette Clarke, DN.Y., is a main member of the energy and trade committee. Since 2019, it has introduced and reintroduced “the DEEPFAKES Accountability Act”.
“We don’t protect the American people as we should,” said Clarke. “What my legislation would simply do is require a digital watermark on any fake Deep manufacturing so that people know and can immediately distinguish that it was generated by artificial intelligence.”
CBS News New York
Next May, The bipartisan “back it down act” takes effectAnd will give websites and online platforms 48 hours to delete false content.
Lawyer Susan Jacobsen specializes in defamation and confidentiality of data in the Brennan Manna Diamond law firm.
“This act will require covered platforms, which are public websites and mobile applications like Facebook, Instagram, Twitter, to establish a process of opinion and withdrawal,” said Jacobsen.
Jacobsen advises victims to capture or screen record videos of any depth and to document the damage he causes. She says that victims can pursue damage and request justice orders for withdrawals if they can identify the perpetrators.
Technological experts show how easy it is to make a deep fake
Rozner spoke to three different technology experts who showed him how easy it was to create an Ai Deepfake video of her.
Matt Swider, editor-in-chief of the Revue Tech website, TheShortcut.com showed Rozner a platform that offers “instant vocal cloning” and another that synchronizes lips.
He took a live report from Times Square and entered a new script, with the version generated by AA it ready in less than an hour.
Rozner noticed that his voice was deeper in the manipulated video.
Swider replied: “If I downloaded more video from you, I would be able to get an exact correspondence, but it was only 20 seconds of video.”
Lionsgate Network helps victims of financial fraud dupped by AI. The founder and CEO of Bezalel Eithan Raviv says that in the last six months, the cases have increased.
“SO [bad actors] Learn from each online interaction, “said Raviv.” It is not a question of hacking a computer or hacking a code. It’s about hacking your conscience. “”
Kwindla Hultman Kramer, the CEO of “Daily”, creates virtual avatars for tutors in education and customer service programs.
Consumers should be targeted in environments they are waiting for themselves.
“The very large social media societies where you would be especially on these videos and in a relaxed observation, you may not look for the signs of a Deepke Deep,” said Kramer.
The fight against Deepfakes “was not a priority”
The FBI says that its internet crime complaints have received more than double the number of complaints concerning videos of the Deepfake AI so far compared to the same period last year.
The agency also has statistics showing that financial losses have almost tripled.
“This is a problem that can be dealt with through the FCC, can be treated through the Ministry of Internal Security, but has not been a priority,” said representative Clarke.
As for deciphering online medical advice, Goldman says: “I always like to tell people, take a break before reacting, right? Stop, ask, ask your doctor. As honestly, don’t send me an email. Ask your health professional, is that something you should take.”
As for how to identify a deep AI, experts say that you should look for clumsy facial movements, lip synchronization problems, a person’s hand and all models of unnatural discourse.
If you do not know the validity of medical approval, experts say they are looking for the doctor’s official website and social media to see if they have announced the approval of their pages.
