

If your company runs Java applications in production, there is a critical security vulnerability that needs attention right now. It is called Java4Shell, tracked as CVE-2026-114938, and it carries a severity score of 9.2 out of 10. For context, that puts it in the same category as Log4Shell, the 2021 vulnerability that sent every security team scrambling.
This one is different in its technical roots but similar in its blast radius.
The bug lives inside the OpenJDK Java Virtual Machine itself. Modern Intel and AMD processors include a set of instructions called AVX-512 VBMI that make certain operations extremely fast. OpenJDK takes advantage of those instructions for Base64 decoding, which is a routine operation that happens constantly in web applications. The optimized code, however, miscalculates how much memory to allocate when it encounters an unexpected number of padding characters in the input.
That miscalculation causes an out-of-bounds write into heap memory. No authentication required.
An attacker who sends a carefully crafted request can overwrite adjacent memory and, with enough effort, execute arbitrary code on the server. Security firm Wiz used an AI-assisted vulnerability research tool to build a working proof-of-concept exploit within two hours of the disclosure. That is a sobering data point.
Affected applications include Jenkins, Atlassian Confluence, Atlassian Jira, Apache Tomcat, Keycloak, JFrog Artifactory, Elasticsearch, and Spring Boot applications, among others. If your team builds or hosts Java applications on any of these platforms, treat this as urgent.
Remediation is straightforward. Upgrade your OpenJDK distribution to the patched version released on November 17, 2026. This covers Amazon Corretto, Eclipse Temurin, Microsoft Build of OpenJDK, Azul Zulu, and other major downstream distributions. If immediate patching is not possible, add the JVM startup flag -XX:-UseBASE64Intrinsics to your application server configuration. This disables the vulnerable code path at some performance cost but eliminates the exposure entirely.
ARM64 processors are not affected. Intel and AMD CPUs manufactured since around 2021, including Ice Lake, Sapphire Rapids, Zen 4, and newer generations, are the vulnerable hardware. If your Java workloads run on modern cloud instances, assume you are on affected hardware.
A few years ago, a vulnerability like this might give security teams two or three weeks before active exploitation appeared in the wild. AI-assisted exploit development has compressed that window dramatically. The window to patch before attacks begin is short.
Prioritize any externally accessible Java application, especially those handling authentication or file uploads. Restrict public internet access to management portals like Jenkins and Jira until patches are confirmed in place. If your organization runs these platforms on behalf of clients, that communication needs to happen today.
The pattern here is worth noticing beyond the specific CVE. AI tools are now speeding up both sides of the security equation. Defenders can find and analyze vulnerabilities faster. Attackers can build exploits faster. The organizations that keep up are the ones with mature patching processes already in place, not scrambling to build them when the fire starts.
Want to explore how proactive security and DevOps practices could protect your business? Let’s talk.

