What is de-identification?

Definition

De-identification means stripping out or swapping the details that identify a person (names, account numbers, addresses, exact dates of birth) so what's left can't reasonably be tied back to them. Our AI policy template defines de-identified information as information from which those details have been removed or replaced so the client can't reasonably be identified from it alone.

In practice, advisors do it by using placeholders like "Client A" and rounding figures.

Why it matters for advisors

De-identifying a question is how many firms let advisors use general AI assistants for planning help without putting client data in the tool. It's simple, fast and lowers risk.

It's also easy to get wrong, because combinations of ordinary details can still point to one person.

It also matters for what you test with. When trialing a new AI tool before due diligence is done, the safe practice is to use made-up or properly de-identified examples, never a real client's file.

Example

Instead of "Jane Smith, 62, $2.4M IRA at Schwab, retiring from Acme in May," the advisor asks about "a client in her early 60s with a seven-figure IRA, retiring next year."

Watch out for

If your firm is small and the details are unusual, even de-identified facts can be recognizable. When in doubt, use an approved tool.