The U.S. Cybersecurity and Infrastructure Security Agency (CISA) has told organizations to patch the Java flaw nicknamed "Java4Shell" immediately, as security teams published the first detailed picture of how the bug works and which machines are actually exposed.
The flaw, CVE-2026-114938, sits in the Base64 decoder of the x86-64 HotSpot Java Virtual Machine. According to CISA, specially crafted Base64 input can crash the JVM and allow unauthenticated remote code execution. Oracle fixed it on November 17, and IntelFusions covered that release when the mechanism was still undisclosed. Canada's Cyber Centre contributed to the CISA alert.
Only newer chips take the vulnerable path
The detail that matters most for triage is hardware. Cisco Talos says the bug affects OpenJDK 17.0.6 and later only on x86-64 processors with AVX-512 VBMI support enabled, a set of vector instructions the JVM uses to speed up Base64 decoding. ARM processors are not affected. Wiz Research names the exposed CPUs as Intel Xeon Ice Lake, Sapphire Rapids, Emerald Rapids and Granite Rapids, plus AMD EPYC and Ryzen chips from Zen 4 onward. Those are common in both data centers and public clouds, so Talos describes the potential attack surface as large.
Wiz says the optimized assembly miscalculates the output bounds when input carries an unexpected number of trailing "=" padding characters. The length calculation underflows to a negative value, and the decoder writes past the end of its heap buffer, which can be used to overwrite adjacent object metadata. Wiz credits the discovery and disclosure to the security firm Calif, which Talos also thanks for sharing details ahead of release.
Wiz says it proved remote exploitation feasible against standard applications including Jenkins, Confluence and Apache Tomcat, and lists Jira, Keycloak, JFrog Artifactory and frameworks built on Spring Boot, Struts, Kafka, Solr and Elasticsearch among applications it found vulnerable before authentication. Its AI-based research harness built a working proof of concept within two hours, although Wiz cautions that generalizing an exploit to arbitrary targets is difficult. Downstream builds are in scope too: Amazon Corretto, Eclipse Temurin, Microsoft's OpenJDK build, Azul Zulu, BellSoft Liberica, SAP SapMachine, Red Hat OpenJDK and Oracle Java SE.
Is anyone exploiting it yet?
Not confirmed. Talos says it has not observed in-the-wild exploitation but expects probing soon, and Sophos X-Ops says no active exploitation had been reported as of its post. GreyNoise has opened a live activity report to track scanning and exploitation attempts; its public summary gives no counts. Scores differ between sources: Talos and Wiz cite 9.2 under CVSS 4.0, while Sophos lists 8.1.
Patch, or start the JVM with Base64 intrinsics off
Apply the November 17 Java updates, or your distribution's patched build, and start with internet-facing services. CISA stresses that Java runtimes bundled inside third-party applications, container images and appliances often do not show up as separate inventory entries, so look inside them too.
To check whether a Linux host runs the vulnerable code path, Wiz suggests confirming that the CPU flags avx512f, avx512bw, avx512vbmi and bmi2 are all present in /proc/cpuinfo. If you cannot patch yet, Wiz recommends adding -XX:-UseBASE64Intrinsics to the JVM startup flags, with -XX:UseAVX=2 as an alternative; both may cost performance. Talos warns that the AVX flag applies only to x86-64, and an ARM JVM will likely fail to start with it. CISA also suggests validating and size-limiting Base64 input at the application boundary and limiting unauthenticated access, and says to watch for unexpected JVM crashes, rogue runtimes and suspicious Java processes.
Talos has shipped Snort rules 1:67502 to 1:67513 (Snort 3 also 1:301754) that flag malformed Base64 in HTTP fields linked to the bug, but Wiz warns that because the flaw lives in the runtime, many code paths can reach it and network rules will never cover them all. The window between a working proof of concept and real attacks is usually short when the target is this widespread.
This briefing is provided by IntelFusions for informational and defensive purposes only. It is based on sources assessed to be reliable at the time of writing, and analytic judgments carry the confidence levels indicated. Indicators of compromise are defanged; re-arm them only in controlled environments. IntelFusions is not affiliated with the organizations named and makes no warranty as to completeness or accuracy.