2026-10-27OPINION · CYBERRISK · BOARDACCOUNTABILITY · RANSOMWARE · RISKREPORTING5 MIN READ READ
FILED UNDER

Your Incident Count Is Flat. That Is the Wrong Number.

ASD's latest data shows fewer high-category incidents and more extensive compromises. Boards tracking volume are watching the wrong chart.

Read the headline numbers in ASD’s Annual Cyber Threat Report 2025–26 and you could come away reassured. ASD’s ACSC recorded no Category 1 incidents and exactly one Category 2 for the year. Ransomware incidents held steady at 139. DoS and DDoS incidents fell by more than 60%.

Read one layer down and the picture inverts. Extensive compromises, the incidents that actually take an organisation apart, rose from 26 to 40. Extensive ransomware incidents rose from 16 to 24, while the total count stayed flat. Extensive compromises from DDoS went from one to seven, while the number of DDoS incidents collapsed.

Fewer incidents are doing more damage. If your board’s cyber dashboard leads with a count, it is reporting the part of the trend that is not moving and hiding the part that is.

The category measures who you are, not how badly you were hurt

ASD categorises every incident on two axes: the effect on the victim, and the significance of the victim to the nation. A federal agency or regulated critical infrastructure operator suffering an extensive incident is a C2. A medium-sized business suffering exactly the same incident is a C4. A small business is a C5.

That is the right design for a national agency triaging its own response. It is the wrong lens for a board reading the report to understand its own exposure. “Only one C2” is a statement about where the damage landed, not how much damage there was.

Here is where it landed. Of the 40 extensive compromises, one hit federal government or regulated critical infrastructure. Thirteen hit state government, academia, large organisations and supply chains. Twenty-six hit small and medium-sized organisations, schools and local government.

The cost data points the same way. The average self-reported cost of cybercrime for a medium business rose 45% to $141,300. For a large business it fell 27% to $148,100. The gap between a mid-market company and a large enterprise, in what a cyber incident costs them, has all but closed. The gap in what they spend preventing one almost certainly has not.

ASD notes that it changed its categorisation method this year, so year-on-year comparisons of categories need care. That caveat cuts both ways. It is a reason to be cautious about the trend, and a stronger reason not to take comfort from the category totals.

Why severity rises while volume falls

The report does not claim a single cause. But three things in its own findings are consistent with the shape.

Ransomware has professionalised. Affiliates buy initial access, use tooling built by specialist developers, and pick targets with the most leverage. A more selective ecosystem produces fewer, deeper intrusions. ASD’s own analysis found a high rate of “Data Encryption for Impact” among C3-and-above incidents and called ransomware a strong contributor to the most serious cases.

Exploitation has also sped up. ASD records a successful exploitation of a vulnerability first identified in 2008, alongside a 30% rise in published CVEs and an explicit warning that AI is compressing time to exploit. When the initial foothold is cheap and fast, attackers spend their effort on going further once inside.

And the low-severity noise is being caught. Unsuccessful low-level incidents make up nearly half of everything ASD recorded. Commodity attacks are mostly being stopped at the edge. What gets through is not commodity.

What the dashboard should show instead

Board cyber reporting is often built around volume, from a time when volume was a reasonable proxy for risk. It leads with incidents detected, phishing emails blocked, vulnerabilities closed. Those numbers measure activity. They say very little about whether the organisation could survive its worst plausible day.

Report by effect, not by count. Ask management to show incidents by how far they got: blocked, contained to one system, spread across systems, disrupted operations. One incident in the last bucket matters more than a thousand in the first, and the dashboard should make that obvious.

Track time to contain, not time to detect. Extensive compromises are the ones that were not contained. The board should see how long it took to isolate affected systems in each material incident and in each exercise, and whether that number is improving.

Ask what “extensive” means for us. For each critical service, management should be able to say what an extensive compromise would take offline, for how long, and what it would cost. If that answer only exists in an incident response plan nobody has tested, it does not exist.

Stop reading national categories as your risk rating. If your organisation is mid-sized, ASD’s scheme will rate your worst day as a C4 or C5. That rating governs ASD’s response priority, not your insurer’s payout, your customers’ patience or your regulator’s view.

Price the middle market honestly. Medium businesses now carry large-enterprise incident costs without large-enterprise security teams. If you are one, your risk appetite statement should reflect that. If you supply one, so should your third-party risk assessment.

The trend is in the tail

Averages and totals smooth out exactly the thing a board most needs to see. The 2025–26 data shows a tail getting heavier while the middle of the distribution stays still. Organisations whose reporting is built around the middle will be told, quarter after quarter, that things are stable, right up until they appear among next year’s extensive compromises.

Next dossier
AI Efficiency Gains Were Liability Transfers in Disguise →
Engage the author
Adam van Vliet is currently taking on briefs.
Brief Adam
Share