Measure 01 · Pressure

How often cyberattacks happen

Attacks rose sharply over the decade. But how big the rise looks depends entirely on how you handle one change the dataset made to its own rules in 2023.

~3×
rise over the decade, adjusted count
~5×
rise if you take the raw count at face value
p ≈ 0.02
trend significant at the 5% level (OLS)
+14%/yr
underlying growth rate, adjusted

Raw vs. rule-adjusted incident counts

In February 2023, EuRepoC began logging every incident against critical-infrastructure targets regardless of who was behind it. The gap between these two lines is how much of the recent surge comes from that wider net rather than more actual attacks.

EuRepoC Global Dataset of Cyber Incidents v1.3.2

Take the raw count at face value and incidents grew more than fivefold between 2015 and 2024, with the single biggest jump between 2022 and 2023, when the number roughly doubled in one year. But most of that jump lines up with the rule change. Strip out the incidents that qualify only because of it, and the adjusted count tells a calmer story: still a clear rise, but roughly threefold, with the 2023–2024 surge mostly flattened. The two lines pull apart sharply from 2023 on — exactly what you'd expect if the rule change, not a real-world shift, drives much of the recent growth.

Is the trend real? OLS fit on the adjusted count

Fitting a straight line to the adjusted counts gives a slope of about +32 incidents per year, significant at the 5% level (p ≈ 0.016), explaining just over half the year-to-year variation (R² ≈ 0.53). The shaded band is the 95% confidence interval — wide, because ten data points is a small sample.

EuRepoC v1.3.2 · n = 10 · OLS with 95% CI

Incidents by sector targeted

Each incident is counted once per sector it hit, no matter how many countries were involved. Watch education climb from almost nothing to a real share by 2023–2024 — attackers widening their aim.

EuRepoC Global Dataset of Cyber Incidents v1.3.2

Incidents by attack type

One incident can carry several type tags, so totals here exceed the incident count. Ransomware runs throughout and rises steeply; no single method drives the whole trend.

EuRepoC Global Dataset of Cyber Incidents v1.3.2

What the numbers can and can't say

Two cautions travel with this. The trend rests on only ten yearly data points, so it's best read as the direction and rough size of the rise, not an exact rate. And the numbers don't climb smoothly — they dip around 2019–2020 before rising steeply — so the trend line describes the decade's overall direction rather than a steady year-on-year march. Neither weakens the core finding of a real upward trend. Both matter for how precisely you state its size.