When the Fideuram fraud hit the news on Friday, my first thought was: I’ve read this one before. A WhatsApp message from an unknown number claiming to be the boss, a confidential deal, a lawyer vouching for it, money that has to move to Asia fast. It was the same script as the attempt against Ferrari in July 2024.
Same script, opposite ending. In Maranello the scam ended with a dropped call. At Fideuram, around 95 million euros went out the door, and part of it still hasn’t come back.
I lined the two stories up side by side, using only what has been published, to see where they diverge.
Ferrari, July 2024#
The account comes from Bloomberg, picked up by Fortune and others. A Ferrari executive gets WhatsApp messages that appear to come from CEO Benedetto Vigna. The number isn’t his usual one, but the profile picture shows Vigna in a suit and tie in front of the prancing horse. The tone is believable: “Hey, did you hear about the big acquisition we’re planning? I could need your help,” followed by a request to be ready to sign a non-disclosure agreement that “our lawyer” would send shortly.
Then comes the call. The voice is cloned and does a good job of Vigna’s southern Italian accent. The fake Vigna explains he’s calling from a different phone because the matter is sensitive, mentions possible China-related snags, and asks for an unspecified currency hedge transaction.
The executive picks up on a few slightly mechanical intonations. Instead of going along, he asks a question: what was the title of the book Vigna had recommended to him a few days earlier? The book was Decalogue of Complexity by Alberto Felice De Toni. On the other end, the call drops abruptly. Ferrari opens an internal investigation and declines to comment.
Fideuram, February 2026#
The sources here are Italian: Il Post, Il Fatto Quotidiano, Forbes Italia, Open and Il Sole 24 Ore, all published on 25 September 2026 based on documents from the Milan prosecutors’ investigation. Some figures differ between outlets, and I flag it where it matters.
On 23 February, Paolo Molesini, then chairman of Fideuram, receives a WhatsApp message from an unknown number. It appears to come from Carlo Messina, CEO of Intesa Sanpaolo, the banking group that owns Fideuram. An urgent international financial operation is needed, and because of technical problems Intesa can’t act directly.
A phone call follows. The voice, cloned with AI, is that of Paolo Nastasi, managing partner of A&O Shearman in Italy, whom Molesini knows and recognises. Nastasi had nothing to do with any of it. Then come emails from addresses that look like the law firm’s, with the destination companies and accounts. Molesini asks Fideuram’s CFO to start the transfers.
Within a few hours, around 95 million euros go out, mostly to China and Hong Kong. The bank’s anti-fraud systems kick in, and a Chinese bank blocks and returns more than 40 million; roughly 13 million more are recovered in Portugal thanks to the prosecutors. Between 36 and 39.5 million are still missing, depending on the source, probably already converted into crypto. On 16 March Molesini resigns for “personal reasons”. According to the press, he is not under investigation.
Side by side#
| Ferrari (2024) | Fideuram (2026) | |
|---|---|---|
| First contact | WhatsApp, not the usual number | WhatsApp, unknown number |
| Who is impersonated | the company’s CEO | the parent group’s CEO |
| Pretext | confidential acquisition, China-related “snags” | urgent international operation, technical problems |
| Third party lending credibility | a lawyer who will send the NDA | a well-known lawyer, with cloned voice and emails |
| Voice channel | cloned voice of the boss | cloned voice of the lawyer |
| Verification | a question about a recent shared fact | no verification outside the attacker’s channels, as far as has been reported |
| Outcome | call dropped, zero euros | about 95 million sent, 36-39.5 million not recovered |
The ingredients are the same: urgency, secrecy, a new number justified by confidentiality, an outside professional acting as guarantor, far-away destinations. AI plays the same role in both cases too. It doesn’t invent anything new. It just makes the step that used to be the scammer’s weak point more convincing: the voice.
Where they diverge#
In the Fideuram case, as far as we know, there was a verification step. The attacker supplied it. The lawyer’s confirming call, the firm’s emails with the account details: every step meant to reassure came through channels controlled by the people running the scam. The confirmation was part of the script.
The Ferrari executive did something different. He moved the conversation onto ground the attacker didn’t control: a private, recent fact that only he and the real Vigna knew. No cloned voice, however good, contains the title of a book recommended in conversation a few days earlier.
It’s the idea behind multi-factor authentication, applied to people. Voice has become a weak factor because it can be copied. A confirmation received on the same channel as the request adds nothing, because whoever controls the channel controls the confirmation too. What holds up is a second, out-of-band factor: something you know, or a channel you pick.
What I don’t want to conclude#
There are two things that bother me in the way these stories get told, and I want to avoid both.
The first is turning the Ferrari executive into a hero and Molesini into a fool. The Ferrari executive had a suspicion, the mechanical intonation, and the suspicion pushed him to check. If the voice had been slightly better, maybe he wouldn’t have asked. A control that depends on one person’s gut works until the attacker does a better job, and AI gets better every year.
The second: I don’t know Fideuram’s internal procedures. I don’t know whether there was dual authorisation for amounts like these, or whether it was bypassed somehow, and the articles I read don’t say. Part of the damage was stopped by the bank’s own anti-fraud systems, so something worked. All I can say is that, in the public account, there’s no check on a channel chosen by the victim anywhere between the first message and the transfers.
There’s also a precedent that makes all this less surprising. In early 2024 an Arup employee in Hong Kong transferred around 25 million dollars after a video call in which every other participant, the CFO included, was a deepfake. If video can already be faked successfully, a voice on the phone is the easy part for scammers now.
The takeaway#
From this comparison I take away three rules, which apply to a company and to a family alike (the “Hi Mum, I’ve got a new number” scam is the same thing on a smaller scale).
- The person receiving the request picks the verification channel. Call back the number you already had in your contacts, never the one the request came from or the one given in the message. The same goes for the lawyer’s contact details.
- Urgency plus secrecy is the red flag. Real requests are sometimes urgent and sometimes confidential. A request that is both, and asks you to skip procedures, should be slowed down on principle.
- Verification belongs in a written procedure. If it depends on someone thinking of it in the moment, sooner or later nobody will. Mandatory callback, dual sign-off above thresholds and for unusual destinations, a phrase or question agreed in advance among the people who can move money. Vigna’s book worked because it was a shared secret that happened by chance; an organisation can create one on purpose.
One question stays open, and I can’t answer it: how long does a shared secret hold up when people’s lives are increasingly on the record? A book recommended in conversation works. A book mentioned in a LinkedIn post doesn’t. My view is that the answer lies in the procedure (a channel chosen by the victim) more than in the question.
Sources#
- Fortune, article on the deepfake attempt against Ferrari, 27 July 2024 (based on Bloomberg, 26 July)
- OECD.AI, Ferrari Executive Thwarts Deepfake Scam Attempt
- Il Post, La truffa da 36 milioni alla banca Fideuram sembra uscita da un film (Italian), 25 September 2026
- Il Fatto Quotidiano, Truffa da 95 milioni a Fideuram: falsi WhatsApp e voci clonate con l’IA (Italian), 25 September 2026
- Forbes Italia, Fideuram, truffa da 95 milioni con WhatsApp e AI (Italian), 25 September 2026
- TrueScreen, Truffa del CEO: il caso Fideuram e come richiamare il bonifico (Italian; timeline citing Open, Il Sole 24 Ore, Sky TG24)
Originally written in Italian, translated with AI assistance and reviewed by the author.