Security for UHNW Family Offices After a EUR 95M Deepfake

Security for UHNW family offices has a new weak point, and it is the most trusted sound in the building: the principal’s voice.

For years, a phone call from the person you work for was the final word. If the principal called the estate manager and said to let someone through the gate, the gate opened. If the principal called the driver and changed the pickup, the driver went. If the principal called the office and asked for the travel schedule to be sent to a new assistant, it was sent. Nobody questioned the voice, because the voice was the authority.

That assumption no longer holds. The case that made it impossible for me to ignore came out of Italy, and it did not involve a family office at all. It involved a private bank, very senior people and a cloned voice. But the chain of trust it exploited is the same one that runs through every family office I have ever worked with.

The broader framework for this article is in High Net Worth Security: The 2026 Guide for UHNW Families.

Security for UHNW family offices when the voice is fake

I am a physical security professional. I protect people, homes and movements. So let me say up front why a fraud case belongs on my desk.

The same technique that can move money can move people and open doors. A convincing call from “the principal” can tell a gate officer to admit a visitor, tell a driver to change the route, tell a nanny to hand a child to someone new, tell the office to share where the family will be next week, or tell household staff to give a courier a package from the safe. Every one of those is a physical security event. Every one of them starts with a voice nobody thinks to question.

Security for UHNW family offices has always been about controlling who can give instructions and who is allowed to act on them. AI voice cloning does not change that principle. It makes it urgent.

What I mean by the trust chain

Every household and every family office runs on a chain of trust. The principal trusts the chief of staff. The chief of staff trusts the estate manager. The estate manager trusts the gate officer. The driver trusts the assistant who sends the schedule. Instructions move down that chain fast, often by text message or a quick call, because speed is the whole point of having good people.

That speed is exactly what an impersonator wants. If you can convincingly sound like someone near the top of the chain, everyone below you will do the work for you.

Key takeaways

• Reporting on the Fideuram case describes a WhatsApp message followed by a phone call using an AI-cloned voice, which led to transfers of about EUR 95 million. Family offices run on the same kind of trust chain.
• The same impersonation that moves money can move people. A fake call from the principal can open a gate, change a driver’s route or release a child to the wrong person.
• Callback rules and two-person approval should apply to physical instructions, not just payments: visitor access, travel changes, child pickups and releases from the safe.
• Code words for principals, family members and key staff are a simple, low-cost control that works when a voice cannot be trusted.
• Recommended hnw security teams test these procedures with realistic scenarios, rather than assuming people will spot a fake.

What happened at Fideuram

According to Biometric Update, citing Reuters, fraudsters using AI to impersonate senior executives stole 95 million euros, about $108 million, from Fideuram, the private banking arm of Intesa Sanpaolo. The scheme reportedly began with a WhatsApp message to the then chair of Fideuram that purported to come from the chief executive of Intesa Sanpaolo, asking for urgent help with an overseas transaction. A follow-up phone call used an AI-cloned voice mimicking a senior lawyer at an associated law firm. Believing the request was genuine, the chair reportedly instructed the finance department to arrange a series of transfers to foreign accounts, mainly in China and Hong Kong.

The same report says authorities recovered about 53 million euros, and that Milan prosecutors placed a foreign national living outside Europe under investigation on suspicion of computer fraud.

The Banker reported that Milan prosecutors are investigating two similar cases, including a 24 million euro theft from specialist lender Banca Ifis and another involving about 2 million euros at an unnamed smaller bank, for a total of about 120 million euros across three lenders.

TechTarget wrote on October 12 that the incident dwarfed previous benchmarks for live executive impersonation.

I want to be careful here. The investigation is ongoing, and the reporting is based on what Reuters and prosecutors have said. I am not adding to that account. What interests me is the structure of the attack, because it is a structure I see in households every day.

These were not careless people

The first thing I notice is who was fooled. This was not a junior clerk clicking on a strange link. It was, according to the reporting, the chair of a major private bank, acting on what he believed was an urgent request from the top of his own group. Senior, experienced, intelligent people are not immune to this. In some ways they are more exposed, because they are used to giving instructions that get followed without friction.

Principals of wealthy families are in the same position. They are used to their word being enough. Their staff are used to acting on it. That is a strength on an ordinary day and a vulnerability on the day someone else borrows the voice.

The WhatsApp to voice call sequence

The sequence matters, so let me walk through it.

First, a written message on a familiar channel, from what looks like a familiar sender. That sets the frame. It tells the target who is supposedly asking and what the situation is. It creates urgency and often secrecy.

Second, a voice call. The call does not need to carry the whole deception. It only needs to confirm what the message already set up. By the time the voice sounds right, the target has already half accepted the story.

Third, action. Someone downstream carries out the instruction because it came from above.

Now apply that to a household.

A text arrives on the estate manager’s phone from a number labeled with the principal’s name, or from a new number explaining that the principal’s phone is broken: “I need a favor, keep this between us. A courier is coming at 4 to pick up the watch box from the study safe. My assistant will call to confirm.” Then a call comes in, and the voice is the principal’s, or close enough. The estate manager wants to help. At 4 o’clock, a stranger is standing at the service gate.

Or: a message to the driver saying plans have changed and the children should be collected from a different entrance at school, followed by a call from someone who sounds like the mother. Or: a message to the nanny saying a family friend will pick up the youngest today.

None of those require hacking a single system. They require a voice, a little background information and a person who wants to do the right thing.

Where the background information comes from

TechTarget made a point I agree with: executives’ public content, including speeches, podcasts, interviews and video posts, can become raw material for deepfake attacks. For UHNW families, that list extends to charity gala videos, social media clips posted by family members, interviews, even a wedding video uploaded by a guest. Audio of the principal is often easy to find.

The background detail that makes the story believable comes from the same place it always has. Staff names on social media. A child’s school in a photo. A travel post. A press release about a property. I cover how those pieces add up in UHNW Protective Intelligence From Online to Physical.

Why family offices share the same trust chain

A family office is, in many ways, a small private bank wrapped around a household. It moves money. It also coordinates homes, aircraft, vehicles, staff, schools, events and travel. Instructions flow in from the principal and the family and out to dozens of people.

I wrote about the cyber controls a family office needs in Cyber Security for Family Offices and UHNW Families, and I will not repeat that here. This article is about the human side: who is allowed to tell whom to do what, and how anyone can be sure the instruction is real.

When I review a family office, I map the instructions that can cause physical harm if they are fake. The list is longer than most people expect:

• Admit a visitor or contractor to a residence.
• Disarm or change an alarm, or share alarm codes.
• Change a driver’s route, pickup point or vehicle.

Release a child from school, an activity or the home to someone.

Share the family’s travel schedule, flight details or hotel.

Remove items from a safe or vault and hand them to a courier.

Book a car or security detail through a new vendor.

Change who has access to the cameras.

For each of those, I ask a simple question: if the instruction came from a perfect copy of the principal’s voice, would it be carried out? In most offices, the honest answer is yes.

Authority has to be designed, not assumed

This is where I bring in the idea of authority control. Every family office should know exactly who can authorize each of those actions, through which channel, and how that authority is confirmed. I go into this in Family Office Authority Control for UHNW Operations. The short version: authority should belong to a procedure, not to a voice.

Callback and two-person rules

The defense against impersonation is old and boring, and it works. TechTarget quoted a former U.S. Attorney recommending trusted channels to confirm transactions, multi-factor authentication and requiring multiple people to approve certain activities. That advice applies just as well to physical instructions as to payments.

Callback on a known number

If an instruction arrives by message or call and it involves any of the high-risk actions on my list, the person receiving it hangs up and calls back on a number they already have. Not the number that just called. Not a number provided in the message. The known number.

If the principal says their phone is broken and they are calling from a new number, that is not a reason to skip the callback. It is a reason to call the chief of staff or another known contact who can confirm.

Principals sometimes find this irritating. I explain it the same way every time: “We are not doubting you. We are protecting you from someone pretending to be you.” Most principals accept that immediately once they understand the point. Some ask why it was not in place before.

Two people for the things that matter most

For the highest-risk actions, I want two people involved. A child is only released to a new person if two authorized adults confirm it. Items leave the safe only with two people present and a confirmed instruction. A visitor not on the approved list is admitted only after the estate manager and the security lead both confirm.

The point is not bureaucracy. The point is that an impersonator now needs to fool two people through two channels, which is much harder.

Urgency and secrecy are red flags

The Fideuram request was reportedly framed as urgent help with an overseas transaction. Urgency and secrecy show up again and again in these schemes: “Do this now.” “Do not tell anyone.” “I will explain later.”

I train staff to treat those words as a reason to slow down. A real principal will survive a two-minute delay for a callback. A fake one depends on there being no delay at all.

Code words for principals and staff

Code words sound old-fashioned. They are also one of the cheapest and most effective controls I know.

I set up two kinds with families.

Verification words

A verification word is a word or phrase that confirms identity. If someone calls claiming to be the principal and asks for something on the high-risk list, the staff member asks for the word. No word, no action, then a callback.

The word should never be written in email or text messages. It should be shared in person, changed on a schedule and changed immediately if anyone thinks it has been exposed.

Duress words

A duress word is different. It is a word the principal or a family member can use to signal that they are being forced to give an instruction. It sounds normal in conversation but tells the listener something is wrong.

This matters for physical security because coercion is real. A principal held at home, in a car or in a hotel room may be forced to call staff and tell them to open the gate, send money or stand down the security officer. The duress word lets the staff member appear to comply while quietly starting the emergency procedure.

Duress words only work if staff know exactly what to do when they hear one, and if everyone practices without making it a game.

Children and code words

For children, I keep it simple. A family password that any adult picking them up must know. Schools and activity leaders are told that no change in pickup is valid without it. If the password is not given, the child stays put and the parent is called on a known number.

Family office high net worth individuals security training

Family office high net worth individuals security depends on the people who answer phones, open gates and drive cars. Those people need training that matches the threat.

TechTarget quoted a security executive making a point I have found to be true: telling people to “be vigilant” does not do much. Walking them through a realistic scenario does. When someone has rehearsed the exact moment a “principal” calls and asks for something unusual, they are far more likely to recognize it when it happens.

What the training covers

My training for family office and household staff covers:

• What the high-risk instructions are.
• How to verify them, step by step.

What to say to a principal who is impatient with verification.

What to do if verification fails.

What a duress word sounds like and what follows.

Who to call, in what order, when something feels wrong.

I keep sessions short and practical. Staff should leave knowing exactly what they will do, not with a vague sense that deepfakes are scary.

Including the principal

The most important person to train is often the principal. If the principal ignores the procedure, or pressures staff to skip it “just this once,” the procedure fails. I spend time with principals explaining why the controls protect them, and I ask them to support staff who follow the rules even when it is inconvenient.

Travel and urgency as pressure

Travel is when these schemes are most likely to work. The principal is away, communication is patchy, time zones are confusing and everyone expects calls from unfamiliar numbers. “My phone died, I am calling from the hotel” sounds perfectly believable.

Travel is also when physical exposure is highest. If an impersonator can get the office to share the hotel, flight or meeting schedule, that information can be used for far worse than fraud.

Protecting itineraries

I treat travel schedules as protected information. They go only to people who need them, through approved channels, and requests to share them with anyone new are verified. A new assistant, a new driver or a new security vendor appearing mid-trip is verified before receiving anything.

Calls from the road

For trips, I agree in advance how the principal will reach the office and the protection team, and which numbers will be used. If a call comes from somewhere else, the verification word and callback still apply. When we provide close protection for UHNW families on the road, our team becomes a trusted channel the office can call to confirm whether the principal really made a request.

Private security for high net worth family homes and impersonation

At the residence, the most likely physical result of impersonation is someone being let in who should not be. That makes the gate, the service entrance and the front door the places where verification really counts.

For private security for high net worth family homes, I want the gate officer and the estate manager working from an approved visitor list. Any addition by phone is verified. Couriers are expected, named and confirmed. Items are not handed to anyone just because someone on the phone said so.

Our residential officers are trained to hold the line politely. “I am sure that is right. I just need to confirm with the office before I can let you in.” Most legitimate visitors are fine with a short wait. Someone who is not legitimate usually reveals themselves by pushing hard.

This is part of home security for high net worth individuals that rarely gets discussed. People invest in gates and cameras, then let a phone call override them.

Personal security for high networth individuals whose voice is public

There is a second side to this that principals rarely think about. Their own voice is now part of their exposure. Personal security for high networth individuals used to mean protecting the person, the home and the schedule. It now also means thinking about how easy it is for a stranger to sound like you.

I am not going to tell anyone to stop giving interviews or speaking at events. Many principals have good reasons to be public. But I do want them to understand the trade-off, and to make a few choices on purpose.

Know where your voice lives

I ask principals and their offices to make a rough list of where long, clean recordings of their voice exist. Podcasts. Conference panels. Earnings calls. Charity videos. Social media clips posted by family members. Voicemail greetings. The point is not to erase them. The point is to accept that a convincing copy is possible, and to stop treating the voice as proof of identity inside the household.

Assume the copy exists, then plan around it

Once a principal accepts that their voice can be copied, the controls in this article stop feeling like an insult to staff and start feeling like common sense. The principal becomes the person who insists on the callback rule, which is exactly what I want. Staff follow procedures far more consistently when the person at the top openly supports them.

Family members are exposed too

The same applies to a spouse, adult children and anyone else whose instructions staff would follow. A teenager’s voice, taken from a video they posted, could be used to call the driver and ask for a pickup somewhere new. That is why private security for high net worth family members includes the verification words and pickup passwords I described above. Everyone whose word carries weight in the household needs to be part of the system.

When the call is a lure

Not every fake call is meant to move money or open a gate. Some are meant to move the principal. A call that sounds like a trusted adviser asking for an urgent meeting at an unfamiliar location, or a voice that sounds like a family member in trouble asking the principal to come immediately, can be used to pull someone away from their protection and into a place an attacker controls.

My rule for principals is simple. If a call asks you to go somewhere unplanned, urgently, without your usual driver or protection, stop and verify through a known number before you move. Tell your security lead or the office where you are going. That habit costs a few minutes and removes one of the easiest ways to set up an abduction.

High net worth security services that test the process

Procedures that are never tested drift. People get comfortable, shortcuts creep in, new staff join without training. The only way to know whether verification works is to test it.

Scenario testing

With the principal’s permission, I run controlled tests. A call to the estate manager asking for a visitor to be admitted. A message to the driver changing a pickup. A request to the office for the travel schedule. The goal is not to catch people out. It is to find the gaps and fix them before someone else finds them.

The results are usually instructive. Staff who would never fall for a phishing email often comply with a confident phone call from someone who knows names and details. That is not a failure of character. It is a sign that the procedure needs reinforcing.

Where in-house research fits

At MSB Protection, we run our own in-house research and development work aimed at countering threats like these. That is a real difference between us and many security providers. For impersonation, it means we keep testing how these schemes are actually carried out against households and how verification, staffing and physical controls can work together so that a convincing voice is not enough to open a door.

What recommended hnw security teams should show you

If you are evaluating recommended hnw security teams or high net worth security services, ask how they handle an instruction from a voice that sounds like the principal. Ask whether their officers would open the gate on a phone call. Ask what their duress procedure is. Ask whether they test it. A good team will have clear, practical answers. A weak one will tell you their people “know the family.”

Knowing the family is exactly what impersonators exploit.

Frequently asked questions

What does security for UHNW family offices include when voices can be cloned?

Security for UHNW family offices should define which instructions carry physical risk, such as admitting visitors, changing travel or releasing children, and require callback on known numbers, verification words and, for the highest-risk actions, two-person approval. Staff need realistic training, and the procedures should be tested.

How does family office high net worth individuals security relate to deepfake fraud?

Family office high net worth individuals security covers the people, homes and movements the office coordinates. Deepfake impersonation can be used to move money, but it can also be used to get access to homes, schedules and family members. The same verification controls protect against both.

Can high net worth security services stop AI voice cloning?

No one can stop someone from creating a fake voice. High net worth security services can make sure no one acts on a fake voice alone, through verification procedures, trained staff, physical access controls and regular testing.

Are code words really useful?

Yes. Verification words and duress words are simple, cheap and work regardless of how good the fake voice is. They have to be shared in person, changed regularly and backed by clear procedures.

How do I choose recommended hnw security teams for impersonation risk?

Ask how their officers handle phone instructions, whether they use verification and duress procedures, how they coordinate with the family office and whether they test the procedures. Look for high net worth security that treats verification as part of daily operations, not a one-time briefing.

Should close protection for UHNW families include verification?

Yes. When we provide close protection for UHNW families, the protection team often becomes a trusted channel. The office can contact the team to confirm whether a request really came from the principal, and the team can use duress procedures if the principal is under pressure.

Conclusion: Make the procedure the authority, not the voice

The Fideuram case, as reported, is a fraud story. For me, it is a warning about trust. A message, a familiar-sounding voice and an urgent request reportedly moved tens of millions of euros. The same pattern, pointed at a household, can open a gate, change a route or put a child in the wrong car.

The fix is not technology alone. It is procedure. Decide which instructions matter. Verify them through known channels. Use code words. Involve two people for the most serious actions. Train staff with real scenarios, include the principal and test it all.

That is how I want security for UHNW family offices to work in 2026, and it is how our residential and executive protection teams operate, backed by our in-house research and development.

For the full picture of how this fits with home security, close protection and travel, read High Net Worth Security: The 2026 Guide for UHNW Families.

Sources

• Biometric Update, Italian bank loses $108M to AI messaging fraud, October 15, 2026
• The Banker, How deepfake fraud is testing bank defences, 2026
• TechTarget, Beyond Arup: CISOs combat escalating deepfake fraud, October 12, 2026

---

About Michael Braun

Michael Braun is a former Special Unit Operator, former Manager at Gavin de Becker & Associates, and Founder & CEO of MSB Protection, an executive protection and residential security firm serving high-net-worth and ultra-high-net-worth clients.

Braun has built his career at the intersection of specialized protective operations, executive protection, residential security, protective intelligence, and security risk management. His experience spans special-unit operations, leadership within Gavin de Becker & Associates, and the development and oversight of private protection programs within demanding UHNW environments.

He has been recognized by The Top 100 Magazine as a leading CEO in the private security field and is the subject of an upcoming Marquis Who’s Who feature highlighting his leadership and contributions to the profession.

Today, Braun is recognized for his work in executive protection, UHNW estate security, residential protection, protective intelligence, adversarial security assessments, and security auditing throughout Beverly Hills and Southern California.

His work focuses on moving private security beyond simply “providing a body” and toward intelligence-led, risk-based protection programs designed to identify vulnerabilities before an adversary can exploit them.

---

Looking for Executive Protection or Residential Security Services?

If you are a high-net-worth or ultra-high-net-worth individual, family office, estate manager, chief of staff, or executive in Beverly Hills, Los Angeles, Malibu or Southern California, MSB Protection provides executive protection, residential security, 24/7 protection, protective intelligence, medical-readiness planning, and security risk management.

We evaluate the complete security environment, from threat exposure and residential vulnerabilities to personnel, technology, procedures, protective intelligence, and emergency response, and build a program around the risks that actually exist.

Contact us for a confidential consultation or message us at +1 (805) 285-2807.