A question on a recent Cyber Daily podcast stayed with me: how many organisations would have named PHP as a crown jewel?

Very few. It is not the CRM, the main website, client data or staff records. It is infrastructure that quietly runs underneath them.

In October, two critical PHP vulnerabilities, now known as PHP2Shell, were disclosed. PHP shipped emergency patches on 21 October, and within hours security vendors were recording thousands of exploitation attempts. The harder question was not "have we patched?" but "do we know where all our PHP is?" It can sit under large websites, and also under small internal tools built years ago and never revisited.

Where "crown jewels" stops

"Crown jewels" is a useful phrase that gives Boards a mental model of what we value most, and I have used it in the education I deliver. Its limitation is that it ranks things by worth, not by what they lean on. An earlier Insight, Before You Protect the Crown Jewels, You Need to Find the Crown, argued that risk only makes sense against the objective it threatens. This piece takes that thinking one layer further down.

The room around the jewels

The real crown jewels sit in specialised security cases on monitored plinths. Visitors look at the jewels. Few think about the plinth, and fewer about what keeps the monitoring alive: the power, the communications link, the alerting tools, and the CCTV systems over the road that the police will use to track down the criminals. Someone needs to think about all of it.

Organisations have the same three layers:

1. Valued. The jewels: key data, top systems, the CRM. Business unit executives know these well.

2. Enabling. The plinth and security systems: the IT architecture, systems and dependencies. The CIO and CISO know this layer.

3. Unseen. What the enabling layer leans on: power, communications, remote access, commodity technology, third-party services and tools nobody remembers deploying. No one holds the whole picture.

Two kinds of unseen

Inherited dependencies are owned by others: suppliers, utilities, hosting and cloud providers, and their own suppliers. The response is assurance. Can they patch promptly, would they tell us, and how would we know? Assurance also covers the incident itself: we often rely on other organisations' controls during and after an event, such as their logs, monitoring and support lines, and those are worth testing before we need them.

Accumulated dependencies are ours, but forgotten. Consider the remote administration tool set up in 2020 so systems could be managed from home during COVID. Meant to be temporary, it became a convenient way to check something, then a habit, then a fixture that never made it onto a register. The response is discovery: looking deliberately for what has been left running, as well as the data we have kept after we no longer need it.

The ownership gap

Business unit executives know their data and top systems. The CIO and CISO know the architecture. Below that, ownership thins out, and the Board's question is not technical: who is accountable for finding what lives in the third layer?

It matters because a great deal can lean on one unremarkable component. One PHP installation might sit beneath the CRM, the public website and an internal tool at once. Its own value is modest. What leans on it is not. Tracing from objectives down, as objective centric risk management does, brings them into view because they appear under many objectives.

Telstra's July outage illustrates the point. A network time server restarted after maintenance with the wrong date, and the fault spread across the national mobile network, disrupting Triple Zero calls. The device was 15 years old and reportedly cost about $30,000. Telstra told a Senate hearing its supplier had warned an update was needed, and that an undocumented configuration change had been made to the unit. It worked quietly until it did not. Only then did anyone see how much leaned on it.

Why speed raises the stakes

The ACSC has alerted organisations to prepare for more frequent updates and shorter timeframes for action. Publicly reported vulnerabilities rose 30 per cent last financial year, and some were exploited within 48 hours of patches or mitigation advice becoming available, with automated and AI-enabled tools helping attackers move faster.

Patching at that pace depends on knowing what we have and who is responsible for it. The inventory, the ownership and our suppliers' ability to patch are themselves part of the plinth. Organisations that know their third layer gain time when the next disclosure arrives.

Putting a process around it

This is a solvable problem. Name an owner for the third layer. Look for what is actually running, not only what the records say, and reconcile the two. Report what is found, and what is still unknown, to the Board or its committee regularly.

Larger organisations may already have enterprise architects, solution architects and third-party teams who can do this. Smaller organisations can reach the same outcome through their IT provider or a periodic external discovery exercise.

One caution applies to both: architecture records describe systems as designed, so the forgotten tool and the undocumented change sit outside them. Discovery against what is really running makes the process worth having.

Three questions for directors

These are questions a Board is well placed to ask, because no single executive owns the answers.

First, what is our appetite for how long critical dependencies stay exposed once a fix exists, and how long they can be offline while we apply it? Directors set those two limits and management reports against them. The more that leans on a dependency, the tighter they should be.

Suppose a critical patch takes eight hours. If exposure is tolerable for a day but the plinth can be offline for only four hours, the gap is resilience, not speed, and the answer lies in investment such as rolling updates or redundancy. If exposure is tolerable for only a few hours, eight hours is too slow for a different reason. If both limits allow it, it is a good answer. The number only means something against the two limits.

Second, can management meet that appetite today, and sustain it? If not, is the gap capacity, capability or investment? The threat has no end date, so a patch process that works this quarter but depends on one person, or a supplier who has not been asked, is not yet a capability.

Third, how would we find out about what we do not know we have, before an attacker does? It asks who is looking for the forgotten tool and the unlisted dependency, and how the Board would know if they stopped.

The jewels will always attract the attention. The Board's contribution is to make sure someone is also looking at the plinth, and at everything it leans on.

This article reflects the author's own analysis, experience, and professional judgement. AI tools were used during drafting to assist with structure, editing, and refinement. The ideas and positions expressed are entirely the author's own.

For more on how Wilk Advisory uses AI, see our AI use statement.

Jason Wilk FAICD is the Managing Director of Wilk Advisory, a governance, risk, and cyber advisory practice based in Perth, Western Australia.

He is a Senior Facilitator and Course Author with the Australian Institute of Company Directors, including authoring all AICD cyber governance courses for board directors.

Contact: jason@wilkadvisory.au

← Back to Insights