Skip to content

What auditors ask about your entropy source, and how to answer

Auditors ask four things about an entropy source: where randomness comes from, what evidence supports the entropy claim, what runs in service to detect degradation, and what records exist for keys already issued. The first three usually have answers; the fourth usually does not.

The questions, the evidence each one requires, and where that evidence has to come from — usable on both sides of the table.

Question one · where does randomness come from

They want: a specific answer naming the source in the system that generates keys, not a general statement about using a secure library.

Evidence: an architecture statement identifying the source, and the path from it to the operation that consumes it. If the answer is the operating system pool, say so — that is a legitimate answer that leads to further questions about the platform.

Question two · what supports the entropy claim

They want: validation evidence with its scope stated.

Evidence: the certificate, its version, laboratory, and exactly which component was in scope — an ESV certificate covers the entropy source, not the module around it. The common failure is a certificate that covers something adjacent to what is being claimed.

Question three · what detects degradation in service

They want: to know the source is still working, not that it worked at evaluation.

Evidence: the health tests that run, their cadence, what they detect and what happens on failure. The sharp version is: what detects a source producing two-thirds of its modelled entropy? If the answer is the mandated continuous tests, the honest answer is nothing does.

Question four · what records exist for keys already issued

They want: to assess exposure for material already in production.

Evidence: health records covering the period in which those keys were generated, retained long enough to be useful. This is where most organisations have nothing at all, and where retention policy decides whether the question is answerable.

Preparing properly

Answer these before being asked. Three of the four are documentation exercises. The fourth is an architecture decision that has to be made before the keys are generated, not after the auditor arrives — which is the whole reason local verification is a procurement argument rather than a technical preference.