Secure Cyber Blog

Java4Shell: How SecureCyber Is Responding and What You Should Do Next

Written by SecureCyber | Nov 18, 2026, 5:41:48 PM

On November 17, Oracle and the OpenJDK project released fixes for CVE-2026-114938, a memory corruption flaw in the Java runtime that researchers are calling Java4Shell. You may also see it called Java2Shell. CISA issued an alert the same day urging organizations to patch immediately.

The flaw sits in the part of Java that decodes Base64, a format applications use constantly for things like logins and file uploads. Specially crafted input can crash the Java process and, in some applications, allow an attacker to run code without logging in. It affects OpenJDK-based Java from 17.0.6 onward, including 21, 25, and 27, on x86-64 processors with AVX-512 VBMI, which covers many recent Intel Xeon and AMD EPYC servers. Standard Java 8 and 11 are not affected.

As of November 18, no exploitation in the wild has been reported. That is good news, but it is not a reason to wait. Java is bundled inside a lot of third-party software, so many organizations are running it in places they have never inventoried.

What we have done and what we are still doing

Our teams started triage the night the news broke, and every department had its assignments before morning. That work has not stopped. The goal is simple: find where our customers are exposed, put eyes on it, and keep them informed as the picture changes.

Network Services pulled the list of every server our customers publish to the internet through their firewalls, and is working through it to confirm deep packet inspection is on for any that are affected. Without it, the firewall cannot see inside the encrypted traffic headed to that server. DevOps is reviewing an external scan for exposed Java. Our SOC built new custom alerts that fire when EDR detects unusual Java activity, with a defined process for investigating each one. Those alerts are live now.

We are also reviewing existing EDR exceptions that involve Java, which we do every time a vulnerability like this drops. Customers are receiving individual reports through their tickets with our findings and recommendations, and we will keep working with them as patches, signatures, and new details are released.

What the firewall can and cannot do

Patching Java is the fix. Network controls reduce exposure and buy time, and that is how we use them.

Our standard FortiGate build blocks all high and critical IPS signatures. When Fortinet releases a signature for an attack like this, it is blocked without a change on our side. For Java4Shell specifically, FortiGuard Labs lists IPS detection as under investigation as of this writing, so we are watching for that update and not counting on it yet.

A web application firewall is the most important network control for a vulnerability like this. Java4Shell is triggered by data sent to an application, and a WAF sits in front of that application and inspects requests before they reach it.

There are two ways to get that protection. The FortiGate includes a built-in WAF profile that covers the basics: signatures for common web attacks, limits on things like request and header size, and control over which HTTP methods are allowed. It is a real step up from no WAF at all, and it is available on the firewall you already own. It is also general-purpose by design, and it can only inspect traffic the firewall is decrypting.

A dedicated WAF goes much further. FortiWeb and the cloud-based FortiAppSec Cloud learn what normal traffic to your application looks like and flag what does not fit, and they add bot mitigation and API protection. FortiWeb also includes a built-in vulnerability scanner. For any application that matters to your business, that is the level of protection we recommend.

For customers who already have a WAF in place, we are reviewing those deployments now. If you publish web applications without one, we can help you choose and deploy the right fit.

The rest of the Fortinet stack helps in different ways. Virtual patching on the FortiGate can shield OT and IoT devices that cannot be updated quickly. FortiClient can tag endpoints based on a detected CVE, which shows us exactly which machines are affected.

This is the second vulnerability of this scale in two months, following PHP2Shell in October. We expect that pace to continue, which is why knowing what you have exposed to the internet matters more than any single patch.

What you should do now

  1. Find your Java. Inventory every Java runtime, including the copies bundled inside third-party applications, containers, and appliances.
  2. Patch. Oracle's fixed builds are 17.0.21.1, 21.0.13.1, 25.0.5.1, and 27.0.1.1. For any other Java distribution, use the fixed build named in that vendor's advisory.
  3. If you cannot patch right away, start Java with the -XX:-UseBASE64Intrinsics option. It is a temporary workaround that turns off the affected code path until you can update.
  4. Start with internet-facing systems. If you publish a Java application through a firewall we manage, tell us so we can confirm it is being inspected.

If you are a SecureCyber customer with questions about your report or your exposure, reply on your ticket or reach out to your account team. Our work on this is ongoing, and we are here to help you through it.