By Samuel Bemi
Most boards do not lack security spending. They lack proof that they could recover, and regulators are about to ask for it.
Picture two board meetings on the same Thursday afternoon. In the first, the cyber item barely registers. The head of IT presents a dashboard of blocked attacks, every indicator is green, and the directors move on. In the second, the chair asks a single question: if our most critical system failed tonight, how long before it is running again, and when did we last prove it?
Both organisations may own the same tools. Only one board is actually governing the risk.
That distinction now matters a great deal. Under the EU’s NIS2 Directive, management bodies must approve and oversee cybersecurity risk measures, and can be held personally responsible when they fail to. The UK’s Cyber Security and Resilience Bill points the same way. Accountability for cyber failure has moved upstairs, whether boards have noticed or not.
The problem: measuring reassurance, not resilience
Budget is the usual explanation for weak cyber governance. It is real, but it misses the point. The deeper issue is that most boards measure the wrong thing, and three habits keep it that way.
First, reporting counts activity rather than outcomes. Blocked attacks, patches applied and tools deployed show how busy the security team has been. They say nothing about how the organisation would cope on its worst day.
Second, accountability drifts downwards. Directors approve the budget, IT runs the tools, and nobody clearly owns the business consequence of a prolonged outage. When everyone is partly responsible, decisions wait.
Third, recovery is assumed rather than proven. A backup schedule exists, so restoration is taken for granted. Until someone actually tries it.
Four disciplines for an accountable board
My career has put me in three different seats around this problem. Early on I designed enterprise networks and ran backup and disaster recovery. Later, as head of IT for a multi-site specialist hospital, I answered to executives for keeping clinical systems available. Since then I have worked in penetration testing and application security, paid to find what others missed.
Drawing on all three, I propose four disciplines for boards that will be held to account: Assume, Measure, Prove, Rehearse. The order is deliberate. Assuming breach changes the question, measurement makes the answer visible, proof makes it credible, and rehearsal tests whether people can act on it under pressure.
Assume. Zero breaches is a promise nobody can keep. I have rarely finished a penetration test without finding something exploitable. A board that plans on the basis of breach asks better questions: which services cannot fail, and how long could each be down before the harm becomes unacceptable?
Measure. Replace activity counts with outcome measures: how long critical weaknesses stay open, how quickly intrusions are detected, and how quickly they are contained. They are rarely flattering at first, which is exactly the point.
Rehearse. A tabletop exercise should test decisions, not familiarity with a document. Who declares the incident? Who authorises emergency recovery spending? If directors are not in the room, the exercise has tested the wrong people.
Proving recovery: where accountability will be decided
Of the four, proving recovery is the one I would put first on any board agenda. Running hospital IT taught me that availability is not a technical metric. When a clinical system goes down, someone is standing at a bedside waiting for it, and you stop treating backups as housekeeping very quickly.
Three rules make the discipline work. A backup only counts once it has been restored under realistic conditions, with the evidence kept. Restore results go to the board with every other material risk, not into an operational pack. And where a system cannot yet be recovered within an acceptable time, that gap becomes an explicit risk decision by a named director, never something quietly left for IT to worry about.
What this does not solve
None of this makes an organisation immune. Some exposures cannot be closed quickly: a legacy system behind an essential service may need funding and a full procurement cycle to replace. Good governance records the exposure, the interim safeguards, the decision-maker and the review date. A well-governed board is not proof of security; it is evidence that the risk is understood and managed deliberately; that is what regulators will look for.
The frameworks already lean this way. ISO 27001 expects risks to have owners and treatment plans, and the four disciplines are a way of acting on that. NIST CSF 2.0, published in 2024, added Govern as a core function, placing cyber risk firmly inside organisational decision-making.
The board’s problem now
Directors do not need to become technologists. The ask is narrower: connect security spending to the services it protects, the people who decide and the evidence that recovery actually works. Assume the breach. Measure the outcome. Prove the recovery. Rehearse the decisions.
Regulators, for their part, should ask to see the last restore test and the last board-level exercise, not just the policy. The chair in that second boardroom was not being awkward. She was doing the job every director is now expected to do.
About the Author

Samuel Bemi is a cybersecurity resilience strategist and subject matter expert in cyber risk governance, recovery assurance and security leadership. He helps boards and executive teams replace reassurance with evidence, tying security investment to the services it protects, clear decision ownership and proven recovery. His work focuses on making cyber accountability practical, measurable and ready for regulatory scrutiny.