Here, we promise to “maximize your earnings in one click with AI”. There, we claim that it identifies stocks whose value will be multiplied by 50. Further down, we praise the merits of artificial intelligence which helps “invest more intelligently and make daily profits”. The messages, illustrated with AI-generated photos of robots and skyrocketing stock prices, use the same arguments: by adopting tools powered by ChatGPT, stock marketers would be guaranteed to win huge sums, very easily.
On social networks, many sponsored ads of this type appear in users’ feeds, without them looking for them. Each time, these messages, which can be seen by tens of thousands of people, hide a scam. Trading scams, as they are called, are a well-known vein that has been exploited for years by criminals. But to convince their victims, scams “surf on the latest developments of the moment”, points out Marc Bouzy, lawyer specializing in cyberscams. A few years ago, cryptocurrencies were at the center of hacker tactics. “It is now artificial intelligence that is in fashion, and we see criminals recycling all the associated vocabulary,” specifies the expert. On
Fake sites, real losses
The scheme consists of putting forward a so-called revolutionary investment technique – using AI, bots, etc. – before redirecting victims to fake sites. Once caught in the process, targets are invited to make an initial deposit on these platforms, then to carry out operations. When they try to make a first withdrawal, this is generally possible, in order to give them confidence. And convince them to deposit even more money.
But these sites are all fake. The operations are not actually carried out. The curves and values displayed are completely invented. These are just shells created by scammers to deceive their victims. And the money they deposited there has long since gone to the criminals’ accounts.
When the pot aux roses is discovered, it is too late. “They never manage to get their money back,” says Jean-Baptiste Boisseau, co-founder of the Signal Arnaques site, which lists and warns about different types of online scams. “For some scams, fake traders call victims directly. In other cases, they try to get them to download malware, or claim that you have to pay fees to withdraw your money,” he lists. One of the latest trends is to use photo montages of celebrities, making it appear that they recommend these trading sites. Eve Gilles (Miss France 2024), Virginie Efira, Elise Lucet and even Yannick Noah have paid the price.
On Signal Scams, anonymous victims share their stories. Like the one who, reassured to see a star promoting a site online promising significant gains, decides to invest 250 dollars and is contacted by a supposed trader promising to find “advantageous investments” thanks to the artificial intelligence. The crook then lends her $5,000 to gain her confidence. “I doubled the stake in two weeks. Too good to be true,” writes the victim. However, faced with the trader’s insistence that she invest larger sums, she decided to close everything.
These scams, particularly devious and complex to set up, are increasingly common on social networks, according to Jean-Baptiste Boisseau. Take a look at Meta’s advertising library is enough to measure the extent of the problem: in December alone, more than 400 active ads leading to questionable sites are visible.
Professional scams
Scammers victimize all segments of the population. The targets are not systematically young, naive people or older people who have little knowledge of social networks. “One of my clients was a very knowledgeable computer scientist, illustrates Anne Bernard-Dussaulx, lawyer specializing in scam cases. The crooks made him believe that they were going to develop together a bot which would improve trading performance. They told him offered to try the one they had already developed, to show him its effectiveness. He lost 200,000 euros.
The criminals go so far as to rob certain victims twice. They can thus be contacted a few months after the theft, by so-called recovery organizations or so-called state services. On the phone, they claim to be able to recover the stolen amounts in exchange for certain fees. “They sometimes even invent taxes that the victims would have to pay,” says Anne Bernard-Dussaulx. Lulled by the promise of seeing their money again, their targets are not suspicious and fall into the trap again.
Once the wrongdoing has been committed, it is very difficult to obtain compensation. It is possible for victims to file a complaint, but the instructions rarely follow through. “In general, these cases are closed without further action, because there are not enough resources for investigations,” sighs Anne Bernard-Dussaulx. The recent “Red Card” affair is an exception. At the end of October, a large-scale trial brought together around twenty people accused of having scammed more than 1,300 victims, for damages estimated at 28 million euros. A vast operation, which targeted both professional football clubs and individuals. The decision taken under advisement will be rendered on March 31, 2025. Sentences of one year to eight years have been requested as well as an arrest warrant against the main suspect, currently in Israel. For years, investigators have regularly seen scams of this type set up by Franco-Israeli networks. Even if she has been improving sincejudicial cooperation in this type of case between the two States has long been insufficient. “Israel very rarely extradites its nationals,” points out Me Bernard-Dussaulx.
Artificial intelligence in trading, a false good idea
Contrary to what scammers claim, generative AIs such as ChatGPT are not relevant tools for traders. “To carry out interesting financial simulations, you have to do econometrics, that is to say take into account all kinds of variables and external factors which will influence this or that event, and then model them,” explains Charles Letillieur, expert in data and artificial intelligence at the consulting firm Converteo. “For these models, it is necessary machine learning capable of spotting trends, not generative artificial intelligence that hallucinates.” In short, ChatGPT is very good at creating and analyzing text. Much less at calculating and analyzing numerical data. “ChatGPT can still help to popularize, explain the most technical elements of trading, or summarize research. But you should not follow his advice when buying,” specifies Charles Letillieur.
Even services that offer automated investment options are very cautious. “Can we really trust artificial intelligence to make wise investments? No,” say the teams at the neo-bank Revolut, which launched an option to make automatic investments in February. “AI has come a long way and is already being used to analyze financial data and identify patterns. But markets are very dynamic, and influenced by factors beyond data and algorithms.” Not to mention that the data on which AI is based may be dated, and therefore less relevant. It is not with ChatGPT that amateur stock marketers will make their fortune.
.