NEW YORK, NEW YORK – MAY 01: Gisele Bündchen attends The 2023 Met Gala Celebrating “Karl Lagerfeld: A Line Of Beauty” at The Metropolitan Museum of Art on May 01, 2023 in New York City. (Photo by Theo Wargo/Getty Images for Karl Lagerfeld)
Getty Images for Karl Lagerfeld
Brazilian police recently arrested four people charging them with using deepfake videos of supermodel Gisele Bundchen in advertisements for “free” anti-wrinkle creams and suitcases for which customers merely had to pay shipping fees. Victims of the scam paid bogus shipping fees but never received the non-existent products. According to police the scammers took in $3.9 million in phony shipping fees.
The advertisements were paid advertisements on Instagram and while Meta stated that it prohibits ads that fraudulently use the images of public figures and that they use AI deepfake detection and trained review teams to identify and take down such ads, they are not particularly effective. The Tech Transparency Project did a study in which it found 663 scam advertisers running more than 150,000 deepfake ads on Meta platforms Facebook and Instagram for which the scammers paid Meta $49 million. Making matters worse, Meta’s ad system uses algorithms that enable precise demographic targeting of ads to susceptible people.
Meta’s actions to remove deepfake ads have been criticized by its own Oversight Board for being slow to flag and remove such ads as well as for being reactive rather than using technology to prevent such ads. Meta disputes the findings of the Oversight Board.
In the United States Section 230 of the Communications Decency Act protects social media platforms from liability for third party content including deepfake ads, however, lawyers have argued that when a platform uses its algorithms to target its advertising to specific people or groups, they actively participate in the post and should not be considered to be protected by Section 230.
Last June, Brazil’s Supreme Court ruled that social media platforms can be held liable in Brazil for illegal content posted by third parties even without a court order.
For years scammers posing as various popular celebrities in ads on social media have lured people into making payments for worthless products. However, with the increased use of Artificial Intelligence and deepfakes these phony ads and scams have become both more believable looking and more frequent as the deepfake and other AI technology became readily available to even relatively technologically unsophisticated scammers.
Tom Hanks warned people of online ads for “miracle cures and wonder drugs” that used AI voice cloning to make it appear as if he was endorsing these bogus products.
A deepfake image of Chef Gordon Ramsey was used to trick people into a scam involving free Hexclad cookware. Videos also appeared on TikTok and Facebook that appeared to show Taylor Swift endorsing and giving away Le Crueset cookware. If you clicked on the link provided you were sent to counterfeit AI created websites of popular cooking websites where you would see further phony testimonials. You were then asked to provide credit or debit card information to cover the cost of the shipping of the free cookware. Unfortunately, just as there was no free Hexclad cookware being given away by Gordon Ramsay, no free Le Crueset cookware was given away by Taylor Swift.
While there are a number of good free and low cost deepfake detection apps, an easy way to determine whether a celebrity ad or endorsement is legitimate is to merely go to the website of the actual product and the celebrity’s website to see if they really do endorse a particular product. Anyone who went to the official websites of both Taylor Swift and of LeCrueset would have learned that the offer of free cookware was a scam.