The Same Method That Faked a Vermeer Now Targets Family Offices

In the 1930s, a Dutch named Han Van Meegeren painted somewhat convincingly forged Vermeers.

Van Meegeren never met his subject. Vermeer had been dead for three centuries. All Van Meegeren had were the outputs. The surviving canvases, and he studied them until he could reason the way the master had reasoned. He wasn’t copying the paintings. He had learned to think and paint like the man who had created them.

The experts authenticated the forgeries anyway. They were guarding provenance. The paper trail of ownership. Not the essence of the work. The forgeries carried none of Vermeer’s literal brushwork and all of his judgment. That was enough.

Last week, the same method possibly surfaced in a more dangerous form. Chinese military labs were reported to be feeding prompts into leading U.S. AI models, collecting the answers, and using those answers to train their own smaller models.

The smaller model inherits the reasoning of the larger one. Then the link is cut. It runs on its own hardware, beyond anyone’s ability to monitor it.

The pattern is simple. Study the outputs, distill the judgment, sever the source, operate free.

Now apply that same pattern to protecting a principal.

A threat building a model of a multigenerational family no longer has to watch them continuously. They only need the pieces the principal, family members, staff, family office, and connected business areas leave in the open. The property record, the aircraft tail number, the staff visible on public networks, the vendor invoices, the announcement that names a venue and a date, just to name a few.

Each fragment seems trivial on its own. Assembled, they form a working copy of a life, and once the copy exists, the watching can stop.

Protective Intelligence must be trained and operate to watch the watchers. The harder activity is looking within and denying the training and operational data to the threat.

The majority of estate environments understanding this internal vulnerability and defend against the information exhaust as well defending the perimeter and gates.

Foresight by Design, not Suprise

Resilience Against Uncertainty