← All posts

Opinion

Late is the new never: what the Cybersecurity Assessment Netherlands 2026 (CSBN) means for SMEs and their MSPs

After the cabinet meeting on 6 November, Dutch Minister of Justice and Security Van Weel responded to the new Cybersecurity Assessment Netherlands 2026 (CSAN, known in Dutch as the Cybersecuritybeeld Nederland or CSBN) with a remark that stayed with me. Some people, he said, believe we need a cyber version of 9/11 to wake us all up. He hoped not.

I hope not either. But I think the alarm has been ringing for quite some time. The National Coordinator for Counterterrorism and Security (NCTV) finished this year’s assessment before Ahold Delhaize was hit in October. According to the NOS, that attack cost the retailer 200 to 250 million euros in revenue, left supermarket shelves empty and kept websites and apps offline for days. We do not lack wake-up calls. We lack people getting out of bed.

This is not only a national picture. In conversations with our MSP partners and their customers, we see the same thing: organisations that know the threat is growing, but are not ready for the speed at which it is growing, a speed that AI is now accelerating.

What does the Cybersecurity Assessment Netherlands 2026 actually say?

The CSBN 2026 makes three points that matter to every organisation, regardless of size.

First, for the first time the NCTV states that attackers demonstrably use AI in cyberattacks. AI helps them find unknown vulnerabilities faster and at greater scale, enables more actors to exploit vulnerabilities in less time, and writes and refines malware. In some cases, malware even consults an AI model during an attack to write new commands or rewrite its own code in real time.

Second, digital resilience is not keeping pace. The report is unusually blunt about this. Despite years of warnings and concrete incidents, part of the Dutch public and private sector still does not have its basic measures in order. The National Cyber Security Centre (NCSC) sees that actively exploited vulnerabilities are still patched too late, despite preventive warnings. None of this is new: the CSBN 2019 already concluded that Dutch digital resilience was insufficient.

Third, the gap is widest for small organisations and SMEs. They face fewer legal obligations, receive less government support and have less knowledge and budget than large companies. The report states plainly that they cannot raise their resilience on their own.

Why is late now the same as never?

Patching late used to be risky but survivable. There was a window between the moment a vulnerability became public and the moment attackers used it at scale. That window has all but closed.

Rapid7’s Q3 2026 Quarterly Threat Landscape Report found that high-severity vulnerabilities were exploited on average three days after public disclosure, with 80 percent of tracked exploitation taking place within 14 days. On the other side of the equation, the Verizon Data Breach Investigations Report 2026 shows that exploitation of vulnerabilities has become the most common initial access vector for the first time in the report’s history. Meanwhile, the median time for organisations to fully remediate a vulnerability listed in CISA’s Known Exploited Vulnerabilities catalogue rose from 32 to 43 days, and only 26 percent of those vulnerabilities were fully remediated, down from 38 percent a year earlier.

These are different datasets, but the direction is unmistakable. Attackers need days; defenders need weeks. And the CSBN makes clear that AI will compress the attacker’s side of that equation further.

This is why I believe the debate about the basics needs a new frame. The basics themselves have not changed: patching, multi-factor authentication, logging, backups, knowing which assets you have. What has changed is the clock. A patch applied after an attacker has been inside for a month is not a basic security measure. It is part of the clean-up.

Why do organisations feel more prepared than they are?

The second uncomfortable message in the CSBN is about perception. The report cites KPN’s Cyberweerbaar Nederland 2026 study, in which organisations rate their own resilience at 7.1 out of 10 on average. Two thirds say they feel prepared for a cyberattack, yet only 28 percent practise this structurally. Only 5 percent prioritise patching, endpoint security and incident response, and 33 percent report having no or only very basic detection. Notably, IT and security staff turn out to be more critical of what their organisation has really arranged than management is.

Bear in mind that three quarters of the respondents in that study work at organisations with more than 150 employees, and 63 percent in sectors that count as critical infrastructure. For SMEs, the CSBN draws on data from Statistics Netherlands (CBS): in 2024, 37 percent of large companies exercised a cyber incident, against 7 percent of small companies and 2 percent of micro-businesses.

There is a deeper problem behind the self-assessment. RAND Europe, commissioned by the NCTV, concluded that there is currently no method that can reliably measure digital resilience. A score of 7.1 is therefore, above all, a feeling.

The CSBN adds one more warning. Both CBS and ABN AMRO report a decline in cyber incidents at companies. That sounds reassuring, but the NCTV points out that these figures only reflect what companies notice and report themselves. As attacks become harder to detect, a falling number may simply mean that fewer incidents are seen. No evidence of a hack is not evidence of no hack. The most dangerous number in cybersecurity may well be the one you report to your board.

If you cannot patch within three days, what can you do?

Here I want to take the objection from many MSPs seriously, because I hear it regularly and it is valid. Patching every system within days is not realistic for most SMEs. Patches need testing, downtime costs money, customers resist maintenance windows, and some legacy systems cannot be patched at all.

But patching is not the only lever. The Dutch government itself showed this in late September. On Friday 25 September, the NCSC confidentially warned organisations about two serious vulnerabilities in Citrix NetScaler. No patch was available yet, and the vulnerabilities were already being actively exploited outside the Netherlands. The NCSC advised organisations to consider disconnecting their Citrix environments from the internet as a precaution. The following day, CIO Rijk, the chief information officer of the Dutch central government, gave central government organisations the line “disconnect, unless” (loskoppelen, tenzij), where the exception applied only if the impact on primary processes would be disproportionate. The patch followed on Sunday evening. Central government did not wait for it; it reduced its exposure.

That is the mindset MSPs can bring to their customers, starting tomorrow and without a single maintenance window.

Do not connect what does not need to be connected. In many SME environments, management interfaces, remote desktop services, old VPN portals, storage devices or cameras are reachable from the internet, often for convenience rather than necessity. A system that cannot be reached cannot be exploited three days after a vulnerability is published. Rapid7’s report gives the same advice: start by identifying externally accessible assets and the vulnerabilities affecting them.

Segment the network. Shipbuilders do not rely on bilge pumps alone; they build watertight compartments, so that one breach does not sink the ship. Separating office systems, servers, production or building technology and guest access means a single compromised device no longer gives an attacker the whole network.

Harden what remains. Disable services, protocols and default accounts that nobody uses. Every component that is switched off is one less component that needs an urgent patch.

None of this replaces patching. What it does is buy time, so that patches can be applied responsibly instead of in a panic. It also fits the “assume breach” principle CIO Rijk invoked after the Citrix incident: plan as if an attacker may already be inside, and limit how far they can get.

Is new technology the answer to AI-driven attacks?

The minister called on organisations to use the latest technology to defend their networks. He has a point. If attackers use AI to find vulnerabilities faster, defenders should use the same capability to find them first. As someone working on AI-driven penetration testing, I see that potential first-hand.

But new technology on top of missing fundamentals only moves the problem. The CSBN cites the Trend Report 2026 of Dutch security firm Hunt & Hackett, which found that the attack techniques in most incidents were well known and detectable; attackers got room because fundamental security measures were missing. The most advanced detection tool sees nothing if there are no logs to analyse.

Who is responsible for SME cyber resilience when SMEs cannot do it alone?

The CSBN is explicit: small organisations and SMEs cannot raise their digital resilience independently, and suppliers, regulators and other government bodies share that responsibility. The Cyberbeveiligingswet, the Dutch implementation of the European NIS2 directive, in force since 15 August 2026, applies to essential and important entities. Most SMEs fall outside its scope.

In practice, this means that for a large part of Dutch SMEs, the MSP is the security department. That is not a burden to resent but a role to claim. It shifts the conversation with customers from availability to exposure: which systems are reachable from the internet, how the network is segmented, what has been hardened, and how quickly the vulnerabilities that matter can be fixed. The customer remains accountable, and its board cannot outsource that. But the MSP is the partner who can turn a feeling of being prepared into actually being prepared.

No cyber 9/11 required

We do not need a cyber version of 9/11. We need fewer open doors on a Monday morning.

AI has not changed what the basics are; it has changed how fast they must be done. Where speed is impossible, exposure has to go down. That is a choice every organisation, and every MSP, can make today.

So here is my question to you: how many of your systems, or your customers’ systems, can be reached from the internet right now? And how many of them really need to be?

Sources