The video is moving, powerful, while the flames wrap the famous Hollywood lyrics. On another, a grandfather pray to Internet users to stay “ten seconds” in his video to save his farm and small animal shoes. In the last one, young women in swimsuit receive sexist insults in the street.
These contents that have social flood networks for a few months have everything to get attention. As proof, some of them combine several hundred thousand views. However, these videos are completely false. They were completely generated thanks to artificial intelligence (AI).
“Even children get there”
The creation of deceptive videos using artificial intelligence is not new. “A few years ago, the production of videos generated by AI was very complicated and, above all, not very credible,” said Tech & Co Victor Fell, a technology teacher. But the arrival of See3, the Google tool known for its hyper realistic images, accelerated the process.
“Today, free and easy -to -use tools have become more democratized. There are still some mistakes in the videos, but they are discreet,” continues the expert. As a result, an army of artisanal influential people produces the deceptive content chain generated by an AI to cheat and extract money from Internet users.
And there is no shortage of examples. Therefore, for a few months, as soon as a natural disaster has occurred in the world, it is quickly accompanied by one or more videos generated by AI. At the end of July, a video of a titanic wave that broke into a Russian coastal resort toured Tiktok … while a real tsunami played the Costa Russian at the same time. The publication exceeded 300,000 visits. The devastating fire videos or tsunamis sprinkling a truck also circulated in mass this summer.
Different scam trends
If these catastrophic contents are a success, they seem to have come from fashion to Tiktok. Because in social networks, even scam trends are ephemeral. “The cameramen are copying among them,” says Victor fell. “They try to see what works more on the platforms and as soon as a video category views, they hurry to violation.”
For a hundred euros per month, some scammers offer mini automated programs, training and even scam trends monitoring to help other criminals produce mass viral content. And they are full of imagination to follow the last lying. At the beginning of summer, the fun videos around the monkeys and others did their businesses. A rabbit sequence that jumps from a springboard has made 72 million visits.
More recently, they were tested with sexist and misogynist videos. As AFP identified, some clips generated by AI claim to show young women in bikini in the streets of India or the United Kingdom. When they “interview” men in the street, then they show sexist or misogynist comments. Here again, these contents are a success. Some videos combine dozens of millions of views.
During the last days, the trend has changed again. Now, cameramen bet on the elderly. The device is always the same. In Melancholy Music, an elderly person, completely generated by AI, asks Internet users to remain in video to save their business. Sometimes these are shoes, other cats for cats or campaign bags to “save cats.”
“My farm will close, but keep it when you are alone 5 seconds,” says a farmer in a video. “The costs accumulate and I can not keep it open,” adds the man, while the images of beautiful animals parade. “We decided to sell shoes to save the farm.”
More than 600 euros per day
Whatever the format, all these apprenticeship scammers try to arouse strong emotions among users, such as fear, pity or even outrage. “It is typical of the economy of attention,” says Tech & Co Stéphanie Laporte. “The emotion generates sharing in networks and the video goes viral quickly.”
And that is the objective of these video creators: “Transform feelings into advertising income,” he adds. In fact, the more a publication generates, the more it will be presented in the algorithm. However, this commitment generates income thanks to advertising and remuneration mechanics for platforms.
“With automation, some cameramen share a dozen content per day. Even if only one of them reaches a million, it is still 600 euros per day,” adds the expert.
Dropshipping, Darknet and false awards
At the same time, “these very followed profiles can be sold to thieves against an orderly sum in Darknet or serve directly as a gate to multiple scams,” says Victor fell. The owners of these accounts use these videos as a call to redirect them to other sites, full of advertising or scams.
This is the case of the famous videos of the elderly in trouble. After having a lot of straw cabins, false entrepreneurs encourage Internet users to buy their crafts to help them straighten their businesses. But it is dropshipping, a practice that consists of reselling more loved articles bought on Chinese platforms.
For example, the sheep shoes of this video are sold 24.9 euros at the Fermecocoon.com site. A search for images invested in products allows you to find these same shoes with only 3 euros. The Refugeininin.com site, for its part, offers bags at 25 euros. Again, the article is sold for less than one euro in Aliexpress.
In disaster videos, cameramen prefer to share prize swimming pools to help allegedly vulnerable people and, ultimately, extract money from Internet users. These doubtful links redirect ads for different scams. Some of the ads hide traps to recover personal data. For their part, the creators of sexist videos often promote a messaging application for adults that allows “making new friends.”
“Everyone can be deceived”
But platforms fight to detect this content. “With the number of videos published per day, algorithms cannot detect everything,” said Ayoub Faouzi. “The platforms have no interest in classifying this content generated by AI,” added Victor Fell. “These are very attractive content that contains users.”
However, the risk is large. In addition to financial damage, Stéphanie Laporte points out the phenomenon of loss of collective trust. “With automation. The production of ‘waste’ in the networks becomes massive. It is increasingly difficult to distinguish the true from the false.”
Especially because “everyone can be deceived,” said Victor Fell. “These more and more sophisticated videos are observed in very small on the screen of smartphones. They are often blurred when even the wisest may not pay attention to the amount of fingers in a strange image or mouth.”
Therefore, experts campaigned for better supervision of this content generated by AI, which should be systematically labeled as such. Meanwhile, it is better to redouble surveillance.
Source: BFM TV
