NewsCryptoReported AI Voice Scam at Italy's Fideuram Sent €36 Million Into Cryptocurrency

Reported AI Voice Scam at Italy's Fideuram Sent €36 Million Into Cryptocurrency

Author: AI Crypto Core·

Key Takeaways

  • •Attackers reportedly used AI-generated voice synthesis to impersonate a trusted individual and issue fraudulent payment instructions to personnel at Fideuram, one of Italy's largest private banking and wealth management groups.
  • •At least €36 million allegedly involved in the scam was reportedly converted into cryptocurrency following the fraudulent payment instruction.
  • •If confirmed, the incident would rank among the largest known AI-assisted voice impersonation frauds against a European banking institution, far exceeding precedents such as a 2019 UK case of roughly €220,000 and a 2024 Hong Kong case of about US$25 million.
  • •The alleged conversion of funds into cryptocurrency could complicate asset recovery because on-chain transfers are irreversible without the cooperation of exchanges or wallet custodians, and cross-border jurisdictional issues add further friction for investigators.
  • •Key details, including the impersonated identity, the timing of the reported call, the staff members who received the instruction, and the destination of the funds, remain unconfirmed pending official statements from Fideuram, Italian regulators, or law enforcement.
Reported AI Voice Scam at Italy's Fideuram Sent €36 Million Into Cryptocurrency

Italian private banking and wealth management group Fideuram — one of Italy's largest wealth managers — was reportedly the target of an AI-powered voice impersonation scam that allegedly led to at least €36 million being converted into cryptocurrency, according to reports circulating in financial and security circles. If confirmed, the incident would rank among the largest known cases of AI-assisted voice impersonation fraud to strike a European banking institution. For scale, widely reported deepfake fraud precedents — a 2019 United Kingdom case in which a cloned voice impersonating a chief executive produced a transfer of roughly €220,000, and a 4 case reported by Hong Kong police in which an employee transferred about US$25 million after a video meeting with deepfaked colleagues — involve sums far below the figure now alleged.

Key points

  • Attackers reportedly used an AI-generated voice to impersonate a trusted individual when contacting Fideuram.
  • At least €36 million is alleged to have been involved in the fraudulent transfer.
  • The funds were reportedly converted into cryptocurrency following the alleged payment instruction.

What the reported Fideuram AI voice scam involved

The alleged impersonation and payment request

According to the reports, attackers allegedly deployed AI-generated voice synthesis to impersonate a known or trusted individual and issue fraudulent payment instructions to Fideuram personnel. The amount reportedly involved is at least €36 million. However, the information available so far does not identify who was allegedly impersonated, which staff members received the instruction, or when the incident is said to have taken place.

The material available for this article does not identify the specific perpetrators, the recipient wallets or platforms involved, or any official findings from Fideuram, Italian financial regulators, or law enforcement. All characterizations of the incident should be treated as unconfirmed pending official disclosure. Any official statement from Fideuram, Italian regulators, or law enforcement is the natural channel through which the missing details — the timing of the reported call, the impersonated identity, and the destination of the funds — could be filled, and in prior crypto fraud cases recovery prospects have tended to hinge on how quickly investigators could trace assets before they were dispersed across wallets and exchanges.

Why the crypto conversion matters

The alleged conversion of the funds into cryptocurrency is a detail that could complicate any potential asset recovery. Once confirmed on-chain, crypto transfers are irreversible without the cooperation of exchanges or wallet custodians, and cross-border jurisdictional issues add further friction for investigators. As previous major crypto fraud cases have repeatedly shown, rapid conversion into digital assets is a common method for moving illicit proceeds beyond the reach of conventional banking freezes.

Voice cloning technology has reached a fidelity threshold at which short audio samples — sometimes drawn from public recordings or corporate videos — can produce convincing real-time impersonations. The alleged use of this capability against a major wealth management institution suggests that threat actors are now directing inference-grade AI tooling at high-value financial targets, not merely consumer-facing phishing campaigns. The exposure is structural in private banking, where large client-directed transfers have long been authorized through direct voice contact with familiar individuals — precisely the form of identification that cloned voices are designed to defeat. Regulators and banking associations across Europe have repeatedly warned institutions about deepfake-enabled fraud in recent years.

How organizations can reduce AI voice impersonation risk

The reported mechanics of the incident — an AI-synthesized voice issuing a large payment instruction — map directly onto a set of controls that financial institutions and corporate treasury teams can implement without waiting for regulatory mandates.

Verify unusual payment instructions through an independent, pre-established channel

Any payment instruction that arrives via voice call, regardless of how familiar the caller sounds, should trigger a callback verification using a phone number stored in an official directory rather than one provided during the call itself. This out-of-band confirmation step breaks the attack chain on which AI voice fraud depends.

Require dual approval and out-of-band confirmation for high-value or changed payment details

Transfers above a defined threshold, and any instruction that modifies existing beneficiary details, should require sign-off from a second authorized party through a separate communication channel. A single voice call, however convincing, should never be sufficient authorization for a transaction in the tens of millions of euros. Policies governing the custody and control of high-value assets are facing increasing regulatory scrutiny precisely because these attack surfaces are widening.

Preserve records and promptly alert security teams, banking partners, and authorities

If a suspicious payment instruction is identified, the priority sequence is to preserve the audio recording and any communication metadata, freeze the outgoing transfer if it has not yet settled, notify the institution's security team and correspondent banks, and report to the relevant national financial intelligence unit. In cases where funds have already moved into cryptocurrency, exchanges operating under anti-money-laundering obligations may be able to flag or freeze associated accounts if contacted quickly with corroborating documentation.

These steps are framed as general risk-reduction guidance and are not intended as claims about the controls in place at Fideuram or about what specifically occurred in the reported incident. The broader trend of AI-enabled fraud targeting financial institutions is well documented; whether this specific case fits that pattern remains to be confirmed through official channels.

Disclaimer: This article is for informational purposes only and does not constitute financial or investment advice. Cryptocurrency and digital asset markets carry significant risk.