AI Deepfake Scam Cost Italy's Fideuram More Than $100 Million
Cybercriminals used AI voice deepfakes and messaging apps to impersonate senior figures and persuade Fideuram executives to authorize overseas transfers, costing the Intesa Sanpaolo-owned bank more than $100 million. About $60.14 million was recovered, while roughly $40.8 million remains missing.
The target was Fideuram – Intesa Sanpaolo Private Banking, a large Italian private-banking and wealth-management group owned by Intesa Sanpaolo. The bank manages about $513.25 billion in assets, employs more than 7,100 private bankers and reported more than $9.85 billion in net profit, according to the report.
Reuters, citing two sources familiar with the matter, said the attack took place in February 2026, when Paolo Molesini was chairman. Molesini received a WhatsApp message that appeared to come from Carlo Messina, Intesa Sanpaolo's chief executive, asking for help with an urgent overseas transaction. Molesini agreed, according to the account cited by TechRadar.
He was later contacted by someone claiming to be a senior partner at a prominent law firm to confirm the instruction. That person used AI tools to create a deepfake of the lawyer's voice, persuading Molesini that the request was authentic.
Molesini then contacted the bank's finance department and arranged a series of transfers to foreign accounts, most of them in mainland China and Hong Kong. The scam was detected soon afterward, and Fideuram moved to block the transaction. With assistance from law enforcement in China, Portugal and Italy, about $60.14 million was recovered. About $40.8 million remains unaccounted for.
Molesini announced a month later, in March 2026, that he was stepping down as president of both Fideuram and Intesa Sanpaolo Private Banking, citing personal reasons. Reuters reported that neither Molesini nor other Fideuram executives are under investigation, while local authorities are investigating a foreign national living outside Europe for potential fraud.
The incident illustrates how business email compromise scams have evolved with AI. In the pre-AI era, fraudsters typically spoofed or broke into executive email accounts, studied finance employees and sent secret wire-transfer instructions. AI tools now make it easier to create deepfake images, cloned voices and even videos, allowing similar attacks to move onto messaging apps, voice calls and video calls.
TechRadar noted a similar case last year in which scammers used AI to create a deepfake of an Italian minister's voice and tricked businessman Massimo Moratti into transferring about $1.13 million to an overseas account. The money was later recovered, according to the report.