Fraudsters cloned a top Italian lawyer’s voice using artificial intelligence and walked away with €95 million from a major financial firm. The elaborate scheme targeted Fideuram, a private banking unit of Intesa Sanpaolo, and ranks among the largest known AI-assisted corporate heists in Europe.
The attack began in February with a WhatsApp message that appeared to come from Intesa Sanpaolo CEO Carlo Messina. It landed on the phone of then-Fideuram Chairman Paolo Molesini. The message pressed for urgent help with a foreign transaction.
A follow-up call sealed the deception. The caller sounded like a senior partner at a prominent law firm, according to sources cited by Reuters. In reality, AI software recreated that lawyer’s voice so convincingly that Molesini treated the request as authentic.
Convinced of the instructions’ legitimacy, Molesini directed Fideuram’s finance team to execute a series of wire transfers. Most of the money flowed to accounts in China and Hong Kong. Company staff eventually spotted anomalies in the payments and notified banks plus law enforcement across multiple countries.
Authorities in Italy, China, and Portugal worked together to recover roughly €53 million. The remaining €36 million vanished. Investigators say the funds moved through a chain of foreign accounts before being converted into cryptocurrency. Neither Intesa Sanpaolo nor Fideuram offered comment to Reuters.
Milan prosecutors have placed one suspect under investigation for computer fraud. That individual, a foreign national living outside Europe, remains the sole target of the probe so far. No Fideuram executives face investigation, sources told Reuters. Molesini stepped down as chairman in March, citing personal reasons.
The episode underscores how quickly voice-cloning tools have turned executive impersonation into a viable weapon. Last year, a similar trick using an AI-generated voice of an Italian defense minister fooled businessman Massimo Moratti into sending nearly €1 million abroad. That money was eventually recovered. As detection lags, more financial executives may find themselves negotiating with machines pretending to be people they trust.
