The invisible physical layer of cybersecurity
There is a dependency underneath your entire cryptographic estate that you cannot see, cannot measure, and have never been asked about in an audit. It is the physical source of the randomness your keys are made from, and the honest answer to how you know it is working is that nobody has ever checked.
For the person who signs off on the security posture rather than the one who implements it.
You have an inventory of your applications. You have a certificate inventory, probably incomplete. You have a supplier assessment process, penetration test reports, and a framework you map controls to. You have, in short, an unusually complete picture of your estate by the standards of a decade ago.
Now: which physical component generated the private key protecting your certificate authority, and what evidence exists that it was working correctly that day?
Why the question has never come up
Because no framework asks it. Walk through the controls you report against and you will find requirements about key length, algorithm selection, key storage, rotation, separation of duties and access control. All sensible. None of them reaches the source of the randomness.
Auditors ask what they are given to ask. The question is absent from the frameworks, so it is absent from the audit, so nobody has ever had to have an answer.
Why it is starting to come up
Two pressures. Post-quantum migration has every organisation opening systems that have not been examined in years, and the examination naturally reaches further down than usual. And supply-chain assurance has established the principle that you should be able to say where a critical artefact came from.
Software has bills of materials. Firmware has signed provenance. Cryptographic key material — arguably the most security-critical artefact you hold — has nothing, and that gap is becoming visible to the people who wrote the other requirements. The question that follows is not one most estates can answer.
What the honest answer is today
For nearly every organisation: we use hardware from a reputable vendor, it holds a certificate, and we have never verified anything beyond that.
That is not negligence, and it is not unusual. It is what was available. The point of stating it plainly is that it is a defensible position right up until somebody asks the question, and then it is not.
The three questions worth asking internally
- For our most critical keys, which physical entropy source produced them, and can we name it?
- What evidence do we have that the source was operating correctly at the time — not that it was certified at some point?
- If it had been degraded, what in our estate would have detected that? If the answer involves statistical testing, it would not have.
Why this is not a scare
The probability that your entropy source is currently degraded is low. The point is not that it is likely; the point is that you have no way to know, and no way to answer if asked.
Every other dependency in your estate has moved from assumption to evidence over the past decade. This one has not, and it is the last one where "the vendor says so" is still considered sufficient. That will change, and it will change during a migration cycle when the systems are already open — which is the cheapest possible moment to have an answer ready.