Critical 'Java4Shell' Java flaw can let attackers run code on Jira, Jenkins
The bug in Java's Base64 decoder hits systems with certain Intel and AMD chips. Oracle has issued patches; no attacks have been seen yet, Talos and Canada's Cyber Centre say.
Key takeaways
- A critical OpenJDK flaw, CVE-2026-114938 or 'Java4Shell', can allow unauthenticated remote code execution.
- It affects Java's hardware-accelerated Base64 decoder on x86-64 chips with AVX-512 VBMI support.
- Calif says it ran code without logging in on Jira, Confluence, Jenkins, Keycloak, Artifactory and PingFederate.
- Oracle patched it on Nov. 17; Talos and Canada's Cyber Centre have not seen exploitation yet.
- Calif rated it 9.2 under CVSS 4.0, while Oracle scored it 8.1 under CVSS 3.1.
A critical memory-corruption flaw in OpenJDK, the open-source version of Java, can let attackers run code remotely without logging in, according to Cisco Talos and the Canadian Centre for Cyber Security; BleepingComputer reports that researchers did so against popular enterprise applications. The bug, tracked as CVE-2026-114938 and nicknamed "Java4Shell," was disclosed on Tuesday, Nov. 17, by the security firm Calif.
Calif scored the flaw 9.2 out of 10 under CVSS 4.0, a critical rating, whereas Oracle gave it a lower score of 8.1, or High, under CVSS 3.1, BleepingComputer reports. Oracle released fixes the same day in its November Critical Security Patch Update, which covers 655 flaws across its products, according to BleepingComputer.
What is the flaw?
The bug sits in Java's standard Base64 decoder, which handles encoded data used in logins, file uploads and network traffic. Specially crafted Base64 input can make the decoder overwrite parts of Java's memory, which could let an attacker steal sensitive data or run malicious code, BleepingComputer reports, citing Calif. The Cyber Centre describes it as a heap-based buffer overflow that could also crash processes.
It affects only a hardware-accelerated decoding routine, BleepingComputer reports, and all three sources say it requires x86-64 processors that support AVX-512 VBMI instructions. Talos says those include commonly deployed Intel and AMD chips, so the potential attack surface is large, in both on-premises and cloud systems. BleepingComputer and Talos say the flaw affects recent OpenJDK 17 updates and later versions, and BleepingComputer adds Java distributions built on that code, such as Amazon Corretto, Azul Zulu, Eclipse Temurin, Oracle Java SE, Red Hat OpenJDK and Microsoft's build of OpenJDK. The Cyber Centre's advisory lists affected Oracle Java SE 8, 17, 21, 25 and 27 builds as well as GraalVM, so administrators should check Oracle's advisory for exact versions.
Which applications are exposed?
Calif says it achieved unauthenticated remote code execution against Jenkins, Jira, Confluence, Keycloak, PingFederate and JFrog Artifactory in their default setups, and authenticated code execution against Apache Kafka, Apache Solr, Elasticsearch, Gerrit and Spinnaker, BleepingComputer reports. There is no universal exploit, the researchers said: attackers must find a way, specific to each application, to feed malicious Base64 data to the vulnerable decoder.
Calif has published a proof of concept showing the memory corruption but has not released its working exploits for the enterprise applications, according to BleepingComputer.
Is it being exploited?
Not yet, as far as defenders know. Talos says it has not seen in-the-wild exploitation but expects probing to start soon after disclosure. The Cyber Centre says it is not aware of active exploitation but warns that this can change quickly.
How was it found?
Calif says researcher Quang Luong found the flaw using OpenAI's Codex coding assistant during an audit of OpenJDK's internals, BleepingComputer reports. According to the researchers, finding the bug took about an hour, a first working exploit took another two hours, and the total AI usage cost under $200.
What should organisations do?
Talos and the Cyber Centre advise applying the updates, and Oracle urges customers to patch as soon as possible, BleepingComputer reports. The Cyber Centre recommends taking an inventory of Java runtimes, prioritising internet-facing systems and checking third-party products and containers that bundle Java. Where patching cannot happen at once, Talos and the Cyber Centre suggest starting the JVM with AVX-512 turned off using the -XX:UseAVX=2 flag; Talos notes this works only on x86-64 systems, and that ARM machines may fail to start with it. Calif also suggests turning off the vulnerable Base64 routine with -XX:-UseBASE64Intrinsics, according to BleepingComputer. The Cyber Centre also suggests isolating unpatched internet-facing apps from public networks, checking whether Linux machines support AVX-512 with the lscpu command, and reviewing logs for signs of compromise. These are stopgaps, not fixes.
Questions readers are asking
What is Java4Shell?
A memory-corruption flaw in Java's Base64 decoder, tracked as CVE-2026-114938, that can let attackers run code remotely on some Java applications without logging in.
Which systems are affected?
Recent OpenJDK 17 updates and later versions, and Java builds based on them, running on x86-64 processors with AVX-512 VBMI support, which include common Intel and AMD chips. The Cyber Centre also lists Oracle Java SE 8 builds; check Oracle's advisory for exact versions.
What should I do?
Apply Oracle's or your Java vendor's update. If you cannot patch right away, Talos and the Cyber Centre suggest disabling AVX-512 in the JVM with -XX:UseAVX=2 on x86-64 systems.
Sources
- #java
- #openjdk
- #vulnerability
- #remote-code-execution
- #patch
Get the day's top AI & tech stories on WhatsApp: Follow the NextFish channel
Get stories like this every morning.
Related
Cybersecurity4 min
FBI says AI tool ARTEX sped some intrusions to admin access in hours
An FBI alert says the open-source framework has been used against at least ten sectors since August. Barracuda and ConnectWise are among the victims named in reports.
Cybersecurity3 min
Optimizely flaw let attackers plant malware on BBC, CNN and Zoom sites
A researcher found malicious code in scripts used by 178 organisations. Optimizely says it fixed the flaw, removed the code and notified affected customers.
Cybersecurity2 min
Google Project Zero will cut its bug disclosure deadline to 30 days
The team says AI has sped up how fast attackers find and exploit flaws. It drops its extra 30-day patch 'soak time' now and moves to 60 days in 2027 and 30 days in 2028.
Cybersecurity4 min
ConnectWise urges ScreenConnect client updates as SCKit targets technicians
Microsoft, Huntress and Blackpoint have analysed the toolkit used against MSPs. It is built to jump to technicians, but researchers say they have not seen that work.