Key Notes
- Calif says AI-assisted research found Java4Shell in one hour and produced a working exploit two hours later.
- Exposure depends on affected Java builds, supported x86-64 hardware and an application path to the vulnerable Base64 decoder.
- Oracle and OpenJDK have published security updates and advisories, while the disclosures do not confirm real-world exploitation.
Security researchers at Calif have uncovered Java4Shell, a serious OpenJDK vulnerability found with help from OpenAI‘s GPT-6 Cyber model. The flaw can expose sensitive memory or allow remote code execution on affected systems, and Oracle has released security updates.
The New York Times reported the discovery on November 17. OpenAI president Greg Brockman separately confirmed GPT-6 Cyber’s role in the finding, linking it to the company’s Patch the Planet initiative.
One Hour to Find the Bug, Two More to Prove It
Calif’s disclosure says researcher Quang Luong took an hour to discover the vulnerability and two more hours to develop the first working exploit. The AI token bill was under $200. Human researchers directed the investigation and built the strategies used to validate and extend the finding.
The account adds a concrete example to the capabilities OpenAI described when it introduced GPT-6 Cyber. The model’s value here was helping a researcher move from a suspected defect to evidence that it could compromise software, a distinction that matters when deciding which bugs need urgent attention.
The Risk Depends on the Runtime and Hardware
Java4Shell, tracked as CVE-2026-114938, sits in OpenJDK’s accelerated Base64 decoder. Base64 represents binary data as text and is used in many ordinary application functions. Calif found memory corruption in the low-level code that accelerates that decoding.
The vulnerable path requires x86-64 processors with AVX-512 VBMI support, according to Calif. ARM64 processors are not affected by this issue. The application must also expose a route to the vulnerable decoder: Java’s presence on a machine does not by itself establish that it can be attacked remotely.
Calif demonstrated attacks without authentication against applications including Jira, Jenkins and Keycloak in its testing. It also stresses that there is no universal exploit: the steps depend on how an application uses Base64 and manages its data.
Java’s extensive use in business makes the discovery significant. Calif cites its use at roughly 90% of Fortune 500 companies, but that is a statement about Java adoption, not a measurement of how many companies have an exploitable Java4Shell deployment.
Patches Are Available, With Different Severity Scores
The OpenJDK Vulnerability Group lists CVE-2026-114938 in its November 17 advisory. Oracle’s corresponding security update identifies affected Oracle Java SE and GraalVM builds and describes the issue as remotely exploitable without authentication.
Calif rates Java4Shell at 9.2 out of 10 under CVSS 4.0. Oracle and OpenJDK publish an 8.1 score under CVSS 3.1. Those figures come from different versions of the severity framework; they should not be presented as the same assessment. Oracle’s entry specifies high attack complexity and potentially high effects on confidentiality, integrity and availability.
Administrators should use their Java supplier’s advisory to identify affected builds and obtain the appropriate update, including for runtimes bundled with enterprise applications. A patch from Oracle does not establish that every downstream distribution or packaged application has already been updated.
The public disclosures reviewed for this article describe research demonstrations and security fixes. They do not confirm real-world exploitation of Java4Shell. That distinction separates evidence that an attack is possible from evidence that attackers are already carrying it out.
Disclaimer: AIstify is an independent media brand owned and operated by NuvexMedia LLC, publishing news, research, and insights on artificial intelligence, emerging technologies, automation, and related industries. NuvexMedia LLC invests in and collaborates with companies across the AI, technology, software, and digital innovation sectors. These relationships do not influence AIstify’s editorial coverage, and the publication maintains full editorial independence to provide accurate, timely, and objective information. © 2026 NuvexMedia LLC. All rights reserved. This content is for informational purposes only and should not be considered legal, tax, investment, financial, or other professional advice.