Security updates will keep coming. Are you ready?

This week alone brought two serious security warnings for software that a large share of websites run on. Weeks like this will become more common. Security updates are no longer something your contractor does once a year. They are an ongoing job, and you need a partner who responds in hours, not weeks.
This week: Nginx and PHP
On 18 November, researchers at Xint (Theori) disclosed NGINX2Root, five critical vulnerabilities in the Nginx web server. The most serious one has been in the HTTP/2 code since 2015 and lets an attacker run their own code on the server without logging in. The fix is in Nginx 1.30.7 and 1.31.9, and a working exploit is already public. As Xint put it: "Finding five remote code execution bugs at once is unusual for nginx."
A day later, today between 14:00 and 14:30 UTC, PHP releases security fixes for all supported versions. The highest severity is critical. WordPress, Drupal, Statamic and Laravel applications all run on PHP.
Both are serious, but neither is an exception. This is the new normal.
There are more vulnerabilities every year
CVE is the public catalogue of known security flaws in software. In 2025, 48,185 new vulnerabilities were added to it, about 20% more than the year before and more than ever.
Part of the reason is that finding flaws is getting faster. AI tools, Theori's Xint among them, scan millions of lines of code in hours and find bugs that stayed hidden for decades. One example is a flaw in PostgreSQL that had been in the code for more than 20 years. That is good news, because defenders are finding the bugs. It also means more security updates, arriving more often.
Attackers are faster than the fixes
In 2018, an average of 63 days passed between the disclosure of a vulnerability and the first attack.
By 2025, according to Mandiant's M-Trends 2026 report, that figure had dropped to minus seven days. On average, attackers are exploiting a vulnerability a week before a fix is available.
When the fix comes out, the clock does not start. It has been running for a while.
No surprise, then, that the Verizon DBIR 2026 found exploiting vulnerabilities to be the most common way a breach starts: 31% of breaches, more than phishing (16%).
Updates do not install themselves
On most servers, security updates do not install themselves. When NGINX2Root came out, Laravel Forge, the PHP server management platform we also use at CNJ, told its users explicitly that it does not update servers automatically. Servers on Ubuntu 20.04 or older no longer receive security fixes at all and need to move to a newer system.
The Nginx flaw sat in the code for ten years. A server nobody has updated since it was set up now carries every flaw found in that time. That is why website maintenance today is first of all about security.
How we handle it at CNJ
We install security updates for the servers and applications we maintain within 24 hours of release at the latest. The same deadline applies to Nginx and to today's PHP release.
We set that deadline because the gap between disclosure and attack keeps shrinking. A week used to be an acceptable response. It no longer is.
What to ask your contractor
If you are not sure how this is handled for your site, start with these four questions:
Who follows security warnings for our server and application?
How quickly do you install a critical security update?
Which version of the operating system, PHP and web server does our site run on, and does it still receive fixes?
How do we find out that an update has been installed?
If you do not get a clear answer to one of them, that is an answer too. We wrote more about checking a contractor in Is your IT contractor ripping you off? and about common mistakes when choosing one in 9 pitfalls when hiring an IT contractor.
If you use AI: do a first check yourself
If you use ChatGPT, Claude or another AI assistant that can open a web page, you can check the basics of your site in a few minutes. Copy the prompt below, fill in your site address and paste it into the assistant with web search turned on.
Prompt to copy:
You are a website security expert. Help me check the website [your site address], which my company owns. Use only publicly available information and normal page visits, with no attacks, vulnerability scanning or load on the server. Check:
whether the site uses HTTPS and the certificate is valid;
which versions of the web server, PHP and content management system the site reveals in its HTTP headers or source code, and whether those versions are still supported;
whether the site sends basic security headers (Strict-Transport-Security, Content-Security-Policy, X-Content-Type-Options);
whether files that should not be public are reachable, for example .env, the .git folder or backups;
whether there are known vulnerabilities for the versions you find, including NGINX2Root and the latest PHP security release.
For each finding, tell me how serious it is and explain it so that someone who is not a developer understands it. Finish with a list of questions I should ask my contractor.
Keep two things in mind when you read the result. AI only sees what is public: it cannot see inside the server, does not know whether the system gets updated and cannot check backups. A site that does not reveal its versions is not necessarily up to date. So the check is a good starting point for a conversation with your contractor, not a replacement for one. And only use the prompt on sites you own.
Security is now a business decision
Cybersecurity is no longer a topic for the IT department alone. In the second half of 2025, Slovenia's national CERT, SI-CERT, handled 2,248 security events and incidents, not counting phishing. Since 19 June 2025, Slovenia's Information Security Act (ZInfV-1) has been in force. It brings the EU NIS2 directive into Slovenian law and, among other things, requires organisations to look after the security of their suppliers.
A website or application is only as secure as the partner who looks after it. If you are not sure who handles security updates for your website and how quickly, get in touch. We will look at what it runs on and tell you what needs fixing.
Frequently asked questions
What is a security update?
A security update is a new version of a piece of software that fixes a known vulnerability. For a website, that means updates to the server's operating system, the web server (Nginx, for example), PHP, the content management system and the libraries the application uses.
Is my website vulnerable to NGINX2Root?
It depends on whether it runs on Nginx and which version. Versions before 1.30.7 and 1.31.9 are affected, although Linux distributions often build fixes into an older version number. The quickest way to find out is to ask your contractor or hosting provider.
How often should a website be updated?
Security updates for critical vulnerabilities should be installed within a few days, ideally within 24 hours. Routine system and library updates make sense at least once a month, and larger upgrades (a new PHP version, for example) should be planned ahead.
Can updates install automatically?
Partly. A server can install operating system security fixes on its own, but larger updates to PHP, the content management system and libraries need testing first, so that the site still works afterwards. That is why automatic updates do not replace regular website maintenance.
Can I check my website's security with AI?
Partly. An AI assistant with web access can check what is publicly visible: HTTPS, security headers, exposed software versions and known vulnerabilities for them. It cannot see inside the server, so the check does not replace the contractor who looks after your site. You will find a prompt for this check in the AI section above.
Key takeaways
This week: five critical Nginx vulnerabilities and a critical PHP release are examples of what will now happen often.
More vulnerabilities: 48,185 were published in 2025, about 20% more than the year before.
Faster attacks: on average, attackers exploit vulnerabilities a week before a fix is available.
Updates are not automatic: on most servers, someone has to install and check security updates.
At CNJ: we install security updates within 24 hours of release at the latest.
A first check with AI: an AI assistant can check what is publicly visible on your site in a few minutes and give you questions for your contractor.
The partner matters: a website's security depends on how reliable the contractor looking after it is.
We write about it. We also build it.
Let's talk about your project
A 30-minute call to understand what you're building and how cvan we help. No brief required.
Prefer email?
info@cnj.si