
The CRITICALSTART® Cyber Research Unit (CRU) is tracking Java4Shell (CVE-2026-114938), a memory-corruption flaw in the Open Java Development Kit (OpenJDK) Base64 decoder. Malformed Base64 input makes an optimized routine write past the end of a heap buffer on x86-64 servers that support Advanced Vector Extensions 512 Vector Byte Manipulation Instructions (AVX-512 VBMI).[1] Because servers decode Base64 in login headers and tokens, the flaw is often reachable without credentials. Calif, which discovered the flaw, demonstrated unauthenticated Remote Code Execution (RCE) in six enterprise applications including Atlassian’s Jira and Confluence, while independent labs have so far reproduced crashes and memory leaks.[1], [2] Oracle and OpenJDK released fixes on November 17, 2026, several downstream builds are still pending, and no exploitation in the wild is confirmed yet. Organizations should find affected Java runtimes on affected hardware, then patch immediately per CISA or apply the startup flag mitigation, starting with internet-facing systems.
Fixes pending for some builds: Red Hat, Microsoft Build of OpenJDK, and BellSoft Liberica fixes were not yet confirmed as of November 18, 2026. Where no fix exists, apply the startup flag mitigations in the Mitigation Strategies section below.
Calif researcher Quang Luong found the flaw through AI-assisted auditing of the Java runtime and reports building a first working exploit within hours, for under $200 in AI costs. Calif coordinated disclosure with Oracle and the OpenJDK project, which published fixes at 1:00 PM Pacific Time on November 17.[1] The trigger was partly public earlier: an unmerged OpenJDK code proposal posted on September 9, 2026, described the same overflow and included a triggering input.[2] CRU assesses with moderate confidence that capable actors could have identified the bug before disclosure. No threat actor has been linked to exploitation at the time of writing.
A characters followed by =, makes an internal length counter go negative. The routine then writes 62 or 63 bytes past the end of the output array. Java rejects the input only after the write.[1]| CVE | CVSS | Description | Relevance |
|---|---|---|---|
| CVE-2026-114938 | 8.1 (v3.1, Oracle) 9.8 (RedHat) |
Heap out-of-bounds write in the HotSpot x86-64 AVX-512 VBMI Base64 decoder | Unauthenticated network attack; memory disclosure, denial of service, and RCE |
Scores diverge mainly on attack complexity; Red Hat’s preliminary Common Vulnerability Scoring System (CVSS) score is 9.8.[2] Calif demonstrated pre-authentication RCE in default configurations of Keycloak, Atlassian Jira, Atlassian Confluence, JFrog Artifactory, Jenkins, and PingFederate, and post-authentication RCE in Apache Kafka, Apache Solr, Elasticsearch, Gerrit, and Spinnaker.[1] Exposure also requires Intel Xeon Ice Lake or newer, or AMD Zen 4 or newer. Standard Java 8 and 11 builds and all ARM64 systems are not affected.[1], [2]
The mapping below, drawn from the MITRE Adversarial Tactics, Techniques, and Common Knowledge (ATT&CK) framework, describes behaviors that are in scope:
| ID | Name | Relevance |
|---|---|---|
| T1595.002 | Active Scanning: Vulnerability Scanning | Java4Shell-shaped probing observed by GreyNoise sensors[4] |
| T1190 | Exploit Public-Facing Application | Pre-authentication RCE through Base64 in login headers and tokens |
| T1203 | Exploitation for Client Execution | Compromise of a Java TLS client connecting to a malicious server |
| T1212 | Exploitation for Credential Access | JavaBleed heap disclosure of keys and tokens |
| T1499.004 | Endpoint Denial of Service: Application or System Exploitation | Unauthenticated JVM crash |
Organizations running affected Java versions should treat Java4Shell as a critical risk, particularly when hosting internet-facing applications such as Jira, Jenkins, Keycloak, and Confluence. Exposure depends on both the Java runtime and the underlying processor, not just the application version. The same container image can be vulnerable on one server but unaffected on another. Denial of service is the most immediate concern, as attackers can trigger JVM crashes through malformed Base64 inputs. Remote code execution requires more targeted exploitation, but successful attacks could compromise enterprise applications, expose sensitive data, and provide access to credentials or service tokens. Importantly, patching does not automatically eliminate the consequences of prior exploitation, as stolen credentials and active sessions may remain valid. Organizations must therefore assess potential compromise and credential exposure alongside software remediation.
To improve your organization’s security posture in light of Java4Shell potential exploitation, CRU recommends the following:
-XX:-UseBASE64Intrinsics to disable the vulnerable Base64 implementation. Alternatively, use -XX:UseAVX=2 on supported x86-64 systems. Verify that the mitigation takes effect and remove the flags after installing fixed builds.IllegalArgumentException errors, JVM crashes, and unusual service restarts. Use the indicators in the Detection and Hunting section below to guide investigations.This table reflects vendor patch availability as of November 18, 2026. Release status may change as vendors publish additional updates. The availability of a newer build does not necessarily confirm that it addresses CVE-2026-114938. Organizations should verify CVE-specific fixes against official vendor security advisories before considering affected systems remediated.
| Distribution | Affected Versions | Fixed Release |
|---|---|---|
| Oracle Java SE 8 Performance Pack | 8u513 | 8u513-perf |
| Oracle Java SE 17 | 17.0.6 through 17.0.21.0.3 | 17.0.21.1 |
| Oracle Java SE 21 | 21.0.13.0.3 and earlier | 21.0.13.1 |
| Oracle Java SE 25 | 25.0.5.0.3 and earlier | 25.0.5.1 |
| Oracle Java SE 27 | 27.0.1.0.3 and earlier | 27.0.1.1 |
| Oracle GraalVM for JDK 17 and 21 | 17.0.21.0.3 and 21.0.13.0.3 | 17.0.21.1 and 21.0.13.1 |
| Oracle GraalVM Enterprise Edition | 21.3.20.0.3 | 21.3.20.1 |
| Amazon Corretto 17, 21, 25, 27 | Earlier builds; Java 17 from 17.0.6 | 17.0.21.1.2, 21.0.13.1.2, 25.0.5.1.2, 27.0.1.1.4 |
| Azul Zulu 17, 21, 25, 27 | Earlier builds; Java 17 from 17.0.6 | 17.70.201, 21.54.201, 25.38.201, 27.30.201 |
| Eclipse Temurin 17, 21, 25, 27 | Earlier builds; Java 17 from 17.0.6 | 17.0.21.1+2, 21.0.13.1+2, 25.0.5.1+2, 27.0.1.1+4 |
| SAP SapMachine 17, 21, 25, 27 | Earlier builds; Java 17 from 17.0.6 | 17.0.21.1, 21.0.13.1, 25.0.5.1, 27.0.1.1 (CVE-specific fix not confirmed by SAP) |
| Red Hat Enterprise Linux 8, 9, 10; Red Hat build of OpenJDK builds | Java 17 (from 17.0.6), 21, and 25 builds | Pending as of November 18, 2026 |
| Microsoft Build of OpenJDK | Java 17 (from 17.0.6), 21, 25, and 27 builds | Verify vendor advisory |
| BellSoft Liberica | Java 17 (from 17.0.6), 21, 25, and 27 builds | Verify vendor advisory |
| Not affected | Standard Java 8 and 11; ARM64 systems; x86-64 without AVX-512 VBMI | Not applicable |
Confirm processor support (Linux; all four flags must appear):
grep -m1 '^flags' /proc/cpuinfo | tr ' ' '\n' | grep -E '^(avx512f|avx512bw|avx512vbmi|bmi2)$'
Confirm the mitigation flag is active:
java -XX:-UseBASE64Intrinsics -XX:+PrintFlagsFinal -version | grep UseBASE64Intrinsics
None of these signs proves compromise, and their absence does not prove a host is safe.[2]
= and whose length is not a multiple of 4, especially in Authorization headersIllegalArgumentException errors from java.util.Base64$Decoderhs_err_pid JVM crash logs or bursts of service restartsThe CRITICALSTART® Cyber Research Unit (CRU) disseminated a brief of the threat as part of the 11-18-2026 Cyber Threat Intelligence (CTI) Update and is providing this Security Advisory as a follow up. CRU will continue to monitor the situation and work closely with the SOC and Security Engineering team to implement any relevant detections and increase vigilance. Critical Start customers are encouraged to reach out to support@criticalstart.com for any further enquiries.
Java4Shell presents a significant risk to organizations running vulnerable Java runtimes on supported x86-64 hardware, particularly those exposing affected applications to unauthenticated network traffic. Although reliable RCE requires application-specific exploitation, the demonstrated memory corruption and relatively straightforward denial-of-service attacks make delayed remediation a serious concern. The September 9 disclosure of a triggering input also raises the possibility that capable threat actors could have identified the flaw before its public disclosure. Organizations should prioritize patching or applying temporary mitigations, validate remediation across embedded runtimes and container deployments, and investigate potentially exposed systems for signs of exploitation. Importantly, successful remediation must also address potential credential and session compromise, as patching alone does not invalidate secrets stolen before the update.
For more threat reports, including OPTIMJACK and Artex, visit Critical Start’s Intel Hub. This advisory was written using the best intelligence available at the time and is subject to change as additional information becomes available.
To learn how Critical Start can help you detect, investigate, and respond to threats faster, connect with a Critical Start team member.
Get Started →© 2026 Critical Start. All rights reserved.
Customers can contact support@criticalstart.com for further enquiries.
