Privacy.Anonymize OpenBI

Personal data in a table's columns found and replaced, redacted, masked or hashed as the table loads, so it never reaches the model.

Group: Privacy. OpenBI’s own: Power Query M has no function of this name.

Syntax

Privacy.Anonymize(table as table, columns as any, optional options as nullable record) as table

Returns

table. The same table, the named columns anonymized as text; the other columns as they were.

Remarks

• Detector "openbi" (the default) finds names, e-mail and IP addresses, Norwegian national identity numbers (fødselsnummer and D-number), phone numbers, account numbers, IBANs and card numbers, each number checked by its check digits. A name is a common first name followed by a capitalized word, or a first name after a greeting. "presidio" adds what Microsoft Presidio's analyzer finds, at PresidioUrl (OpenBI Studio's Privacy setting unless given) in Language ("en").
• Operator "replace" (the default) writes the kind in its place: <EMAIL>, <PERSON>, <PHONE>, <CARD>, <IBAN>, <ACCOUNT>, <NO_ID>, <IP>, <LOCATION>. "redact" removes it; "mask" turns its letters and digits into * but the last two; "hash" writes the SHA-256 of Salt and the value, 64 hexadecimal digits: the same value always gets the same hash, so joins and counts still work.
• Entities limits it to some kinds, as {"EMAIL", "PHONE"}. Whole = true treats each whole value as personal (an e-mail or customer column), named by Type (PII unless given).
• It runs as part of the table's SQL, so a refresh reads the personal data only to anonymize it. When Presidio is asked for and doesn't answer, the load fails: nothing goes through un-anonymized.
• Detection follows rules, check digits and lists of names: a name it has never seen, written alone, can slip through. Presidio finds more. In Transform data, Home › Anonymize writes this step.

Compared with Power Query

An OpenBI function: Power Query has none like it, so a query that uses it runs in OpenBI only, not in Power BI Desktop or the Power BI service.

Examples

Names, phone numbers and e-mail addresses in comments

Privacy.Anonymize(
    #table({"Ticket", "Comment"}, {
        {1, "Call Emma Hansen on +47 912 34 567"},
        {2, "Mail sent to kari@example.no"},
        {3, "Printer jammed again"}}),
    {"Comment"})

Masked, and only e-mail addresses redacted

let
    Tickets = #table({"Ticket", "Comment"}, {{1, "Call Emma Hansen on +47 912 34 567"}, {2, "Mail sent to kari@example.no"}}),
    Masked = Privacy.Anonymize(Tickets, {"Comment"}, [Operator = "mask"]),
    Redacted = Privacy.Anonymize(Tickets, {"Comment"}, [Entities = {"EMAIL"}, Operator = "redact"])
in
    Table.Combine({Masked, Redacted})

A customer column hashed: counts still add up

Each e-mail address becomes the same hash every time, so the orders still group by customer.

let
    Orders = #table({"Customer", "Amount"}, {{"kari@example.no", 120}, {"ola@example.no", 80}, {"kari@example.no", 45}}),
    Hashed = Privacy.Anonymize(Orders, "Customer", [Whole = true, Operator = "hash", Salt = "shop-2026"]),
    ByCustomer = Table.Group(Hashed, {"Customer"}, {{"Orders", Table.RowCount, Int64.Type}, {"Amount", each List.Sum([Amount]), type number}})
in
    ByCustomer[[Orders], [Amount]]

Related functions

Table.TransformColumns, Table.Group, Text.Replace.

Last updated 8 November 2026 · OpenBI Studio 0.3.79