If it feels like there is a new “critical” security warning in the news every week, you are not imagining it. In the last month alone, two of the most widely used pieces of web software in the world, PHP and NGINX, each had flaws serious enough that an anonymous stranger could take over a server. A big reason for the pace is artificial intelligence: the same tools that help us build software faster are now finding the holes in it faster too.
Here is what changed, what it means for your website or app, and how we protect the sites and software we build and host.
- What Changed This Fall
- Why It Matters for a Normal Business
- How We Stay Ahead
- Security Is a Full-Time Job Here
- What You Can Do Today
What Changed This Fall
Four stories from the last few weeks show the shift:
- php2shell (October 21). PHP, the language behind WordPress and a large share of the web, patched two critical flaws that let an attacker run code on a server without logging in. Within days, researchers confirmed attackers were using them successfully. We wrote a plain-English guide to php2shell at the time.
- AI finding flaws at scale (November 6). OpenAI said its new security model found 81 previously unknown high-severity flaws in Firefox, a browser that has been tested by experts for twenty years.
- Attackers using AI on their own (November 16). Google’s threat intelligence team reported its first case of an AI system finding and exploiting flaws with no human in the loop. Google also counted 22 zero-day attacks in August 2026 alone, up from an average of 8 a month in 2025.
- NGINX2Root (November 18). Two critical flaws were disclosed in NGINX, the web server that sits in front of a huge portion of the internet. One had been hiding in the code since 2015. The researchers at Wiz found both with an AI code-scanning tool, and summed it up well: “The flaws were already there; AI helped us find them.”
Why It Matters for a Normal Business
For years, the main thing protecting most small and mid-size businesses was that finding a serious flaw took a skilled person a long time. That gap is closing. When an AI can read millions of lines of code in an afternoon, old bugs come to light faster, and attackers can turn a published fix into a working attack in days instead of months.
The practical result: the window between “a fix is announced” and “someone is using this against websites” is now very short. A site that gets updated “when someone gets around to it” is a site that will eventually be on the wrong side of that window. Cyber insurers know it too, which is why many now scan their customers’ websites and send urgent alerts when they spot exposed software.
How We Stay Ahead
We take security seriously because our clients trust us with their websites, their customer data and, in many cases, their whole business online. Here is how we handle it, without giving away the details an attacker would want:
- We watch around the clock. Every server we manage is monitored continuously for signs of intrusion: unexpected logins, suspicious files, unusual changes and new openings to the internet.
- We track every advisory that affects the software we run. When a vendor, a researcher or a government agency publishes a new flaw, we match it against the exact software and versions on our servers, so we know quickly whether it affects us.
- We act before the patch when we have to. Sometimes a flaw is announced before the fix is ready. When NGINX2Root was disclosed, the vendor’s patch was not yet available, so we applied the published safeguard across every server we manage the next morning and confirmed it from the outside. Sites kept running normally the whole time.
- Every code change is reviewed for security before it goes live. Missing permission checks, unsafe input and private data shown to the wrong person are caught in review, not in production.
- We test our own software the way an attacker would. We regularly check the apps and sites we build for broken access controls, exposed admin pages and leaked keys, and fix what we find.
Security Is a Full-Time Job Here
As the pace picked up, we made security a dedicated, full-time role on our team, not something that only gets checked during a project review.
The security watch is worked every day. When something looks off, or a new advisory touches software we run, it is investigated, confirmed or ruled out, and a clear plan for the fix is laid out the same day. This week we expanded that role so new advisories for the software on our servers are reviewed the day they are published.
Fixes still go through our normal, reviewed process, so speed never comes at the cost of care.
What You Can Do Today
- Know what your website runs on. Ask whoever hosts your site: what web server, what version of PHP or other languages, and who applies security updates.
- Ask how fast updates happen. “Within days of a critical fix” is the right answer now. “Once a quarter” is not.
- Do not ignore an insurer’s alert. If your cyber insurer emails you about your website, forward it to your web team or hosting provider right away.
- Put someone in charge. If nobody owns updates and monitoring for your site, that is the gap to close first. Our Care Plans cover hosting, updates and monitoring for exactly this reason.
If you are not sure where your website stands, get in touch. We will tell you plainly what we find.
