What did CISA’s 2024-2025 Vulnerability Review find?

Its central finding is that most successful compromises exploit known, already-catalogued vulnerabilities — many on CISA’s Known Exploited Vulnerabilities (KEV) catalog — on internet-facing systems, rather than novel zero-days. Perimeter and edge devices (VPNs, firewalls, file-transfer appliances) are disproportionately targeted, and the dominant weakness classes remain input validation, memory safety, access control, and injection.

How likely is a ransomware attack, and what does one cost?

The review cites Cyentia Institute IRIS data putting the annual likelihood of a ransomware attack at roughly 10% per organisation, with an average total financial loss near $3.7M per incident (downtime, recovery, extortion and regulatory costs — not just the ransom). These are US-weighted figures, useful as an industry sanity-check anchor rather than a Europe-specific rate.

Why does this favour external, telemetry-based cyber risk quantification?

If most breaches come from known vulnerabilities on exposed systems, the highest-value thing you can do is find your own externally-visible weaknesses and weight them by what is actively being exploited (KEV, EPSS). Nisura scans the external attack surface an attacker actually sees and prices the exposure natively in euros against EU wages, margins, and NIS2/GDPR/DORA liability.

CISA's 2024–2025 Vulnerability Review: Attackers Exploit Your Known, Exposed Weaknesses — Not Zero-Days

In August 2026 CISA published its Vulnerability Review for Fiscal Years 2024 and 2025 (TLP:CLEAR). Strip away the agency prose and it makes one blunt point, again and again: the attacks that actually cause damage do not begin with an exotic zero-day. They begin with a known, already-catalogued vulnerability sitting on an internet-facing system that the defender never got around to patching. That single finding is the entire premise of external, telemetry-based cyber risk quantification — so it is worth reading closely.

~10%/yr

chance of a ransomware attack per organisation (Cyentia IRIS, cited in the review)

$3.7M

average total financial loss per ransomware incident (Cyentia IRIS)

Edge-first

VPNs, firewalls and file-transfer appliances are the review's most-cited targets

The headline: known beats novel, every time

The review's central message is that most successful compromises exploit vulnerabilities that were already public and already fixable — many of them on CISA's own Known Exploited Vulnerabilities (KEV) catalog. Zero-days exist and matter, but they are the exception. The rule is an unpatched appliance, an exposed admin panel, a forgotten subdomain still running a vulnerable version. Attackers are economically rational: they reach for the cheapest working key, and a published CVE with a public exploit is the cheapest key there is.

This is exactly why Nisura's engine is built on live exploit intelligence rather than a static scanner signature list. Every simulation is priced against the current KEV catalog and EPSS exploitation-probability scores — the same "what is actually being exploited right now" signal the review says defenders should prioritise.

Edge devices are the front door

The review repeatedly names perimeter and edge technology — VPN concentrators, firewalls, and managed file-transfer appliances — as disproportionately targeted. The logic is grim but simple: these boxes are internet-facing by design, they terminate credentials, and a single unpatched one is a straight line into the network. An organisation can run a mature internal patch programme and still be breached through the one appliance nobody owns.

That is why Nisura scans the external attack surface an attacker actually sees — subdomains, exposed services, appliance versions — rather than trusting an internal asset list to be complete. What you can't see, you can't patch; what you can't patch, an attacker will find.

The recurring weakness classes

On the engineering side, the review's top weakness categories are the familiar ones: improper input validation, memory-safety defects, broken access control, and injection. None of these are new, and that is the point. Decades-old bug classes still dominate because the incentive to fix them cheaply has never overtaken the incentive to ship. The review also flags an emerging shift: AI-assisted vulnerability discovery, which lowers the cost of finding these flaws for attacker and defender alike — and compresses the window between disclosure and exploitation.

What the numbers mean for a euro figure

The review cites Cyentia Institute IRIS data putting the annual likelihood of a ransomware attack at roughly 10% per organisation, with an average total loss near $3.7M per incident — the full cost of downtime, recovery, extortion and regulatory fallout, not just the ransom paid. Those are US-weighted figures, so we treat them as a sanity-check anchor, not the driver: Nisura's headline number is computed natively in euros against EU wages, EU margins, and EU regulatory liability (NIS2, GDPR, DORA). When your own modelled likelihood lands near that ~10% industry anchor, that is corroboration — not coincidence.

The through-line from a US federal vulnerability review to a European CFO's risk register is short: the same known, exposed weaknesses that CISA says cause most intrusions map directly onto NIS2 and GDPR liability for EU companies. Quantifying that exposure in money — before an attacker prices it for you — is the whole job.

The takeaway

CISA's data validates a specific, unglamorous discipline: find your own known, exposed vulnerabilities from the outside, weight them by what is actually being exploited, and translate the result into a figure your board can act on. That is not a novel idea — it is the idea the attackers have already priced in. See how the leading CRQ tools compare, or run your own external exposure through Nisura to get the euro number.