Insider risk · 3.5

Privileged access, and who reviews it

Privileged access, and who reviews it. What the requirement says, what it means in practice, and what an assessor will ask.

For an independent reference point, see CISA Insider Threat Mitigation Guide.

The problem

Administrators can bypass most of what you built

Controls are configured by somebody, and that somebody can generally reconfigure them, clear the record of having done so, and reach anything the system holds.

This is not a reason for suspicion of administrators. It is the reason the standard treats privileged access separately, and the reason a programme that ignores it has a gap running underneath every other control.

Four arrangements

Ordinary practice, in order of effort

Separate accounts. An administrator's daily account is not their privileged one. Cheap, immediate, and it means a compromised email session does not carry administrative rights.

Least privilege. Rights scoped to what the role requires. Most environments grant broad rights because it was simpler at setup and nobody revisited it.

No shared administrative credentials. Named accounts, so that an action attributes to a person. Where a shared account must exist, it is controlled as below.

Logging elsewhere. Privileged actions recorded to a system the administrator cannot alter. This is the one that makes the others verifiable.

Break-glass

The account for when everything else fails

Most environments need one. It should be documented, its credentials held securely and split where practical, its use alerted on, and every use reviewed afterwards.

An emergency account with no alerting is an unmonitored master key, and its existence is usually known to more people than the register suggests.

The small supplier problem

One person who is everything

Separation of duties assumes several people. A manufacturer with one information technology person cannot separate anything, and pretending otherwise produces a document that describes a fiction.

What works instead is compensating: log privileged activity to somewhere that person does not administer, have a second person, not necessarily technical, review the log periodically against a simple question, and require the compliance owner or a director to approve changes to security-relevant configuration.

None of that is as good as separation and all of it is defensible, documented, to an assessor who understands the size of the organisation.

Vendors with privilege

The widest access is frequently external

Managed service providers, equipment vendors and integrators often hold administrative rights, sometimes broader than anybody internal. The entry on remote access covers the route; the privilege itself needs the same treatment as an employee's: named, scoped, logged, reviewed.

The review

Quarterly, and it always finds something

Who holds privileged access, on what, and why. Compared against the list from last quarter.

The recurring findings are an account belonging to somebody who changed role, a vendor account from a finished project, and a rights grant made for a migration two years ago.

Evidence

What to be able to show

The list of privileged accounts with their justification. The separation between daily and privileged accounts. The log destination and who can alter it. The break-glass procedure and its use records. And the quarterly reviews, dated.

Documenting the exception

Where separation is impossible, say so

A written statement that the organisation is too small to separate these duties, naming the compensating controls and who reviews them, is a legitimate answer.

An unwritten one looks like an oversight, and the difference between the two is a paragraph.

Also

Elsewhere in insider risk