Security firm Calif disclosed on Tuesday, November 17, Java4Shell (CVE-2026-114938), a critical memory-corruption flaw in the Base64 decoder of OpenJDK, the world's most widely used Java runtime. Calif rates the flaw 9.2 out of 10 (Critical) under CVSS 4.0; Oracle, which assigned the CVE, rates it 8.1 under CVSS 3.1. The company announced the discovery in a post on X, after the vulnerability leaked ahead of coordinated disclosure, and published the full technical write-up once Oracle's patch shipped at 1 p.m. Pacific.
What the flaw is
To decode Base64 faster, OpenJDK uses a hand-written x86-64 assembly path with the processor's AVX-512 VBMI instructions. Calif found an error in the write-mask calculation for the padded tail of the input: a malformed input of just 65 characters — 64 letter “A”s followed by a single “=” — makes the decoder write 62 bytes past the end of the output array, corrupting neighboring memory in the JVM heap. The flaw affects OpenJDK 17.0.6 and later, including Java 21, 25 and 27, and derived distributions such as Amazon Corretto, Eclipse Temurin, Microsoft Build of OpenJDK, Azul Zulu, BellSoft Liberica, SAP SapMachine, Red Hat OpenJDK, Oracle Java SE and Oracle GraalVM for JDK. It only exists on x86-64 processors with AVX-512 VBMI — Intel Xeon from Ice Lake onward and AMD EPYC and Ryzen from Zen 4 onward. ARM64 processors are not affected.
Who is exposed
In its own testing, Calif demonstrated unauthenticated remote code execution, in default configurations, against Keycloak, Atlassian Jira, Atlassian Confluence, JFrog Artifactory, Jenkins and PingFederate. With authentication, it also achieved RCE against Apache Kafka Broker, Apache Solr, Elasticsearch, Gerrit and Netflix Spinnaker, as well as sample applications built on Apache Struts and Spring Boot. Because Base64 appears in authentication, file uploads and network messages, ordinary features can expose the vulnerable decoder. Calif also described a memory-disclosure variant, nicknamed JavaBleed, that can read heap contents — potentially personal data, private keys and tokens — in the style of Heartbleed.
Discovered with AI assistance
Calif researcher Quang Luong used OpenAI's Codex to audit HotSpot's JIT compiler, intrinsics and cryptography. According to the company, discovery took one hour, the first working exploit two more, and the total token cost stayed below US$ 200. Calif calls this combination of human researcher and AI the “centaur hacker”. There is no universal exploit: each application needs its own steps, and the firm is withholding the per-product exploit chains while users upgrade. As of November 17, no exploitation in the wild had been reported.
Fix already available
Oracle released fixed builds on November 17 — 17.0.21.1, 21.0.13.1, 25.0.5.1 and 27.0.1.1 — and Amazon published fixed Corretto builds for the same lines. CISA, the United States cybersecurity agency, issued an alert urging immediate patching. Until the patch is applied, two JVM options mitigate the flaw: -XX:-UseBASE64Intrinsics and -XX:UseAVX=2 (the latter only on x86-64, as it prevents startup on ARM). After updating, every JVM must be restarted so the fixed build is the one running.
Sources
- Calif — Java4Shell: Write once, RCE everywhere
- Censys — Nov 17 Advisory: OpenJDK Remote Code Execution (Java4Shell)
- BleepingComputer — Critical Java4Shell flaw allows RCE attacks on major Java applications
- CISA — Critical Java Vulnerability Requires Immediate Patching
- JFrog Security Research — Java4Shell: What We Know About the Critical OpenJDK RCE Vulnerability
- Datadog Security Labs — Java4Shell: Remote Code Execution in OpenJDK
Image credit: Calif — illustration of the “centaur hacker” from the Java4Shell disclosure write-up.
Transparency: this article was produced with the assistance of artificial intelligence, based on cross-referencing public posts on X and internet sources. Consult the original sources to confirm the information.
Por Muse