Thursday, November 12, 2026America’s Trusted News SourceFollow: X · Facebook · Bluesky

USA Pulse News

BREAKING · POLITICS · WORLD · SPORTS
LIVE
BREAKING
News

AI Finds Critical Java Flaw in One Hour as Oracle Rushes Out Emergency Patch

Advertisement
Advertisement
Lines of Java code on a computer screen. The Java4Shell flaw sits in OpenJDK's Base64 decoder. (via fit-lounge.at)

Security researchers have uncovered a critical flaw in Java, one of the world's most widely used programming platforms, with the help of artificial intelligence. The vulnerability, dubbed Java4Shell and tracked as CVE-2026-114938, is a memory-corruption flaw in the hardware-accelerated Base64 decoder of OpenJDK, the open-source implementation of Java that powers enterprise applications, banks, hospitals, airlines and government systems.

Researchers at cybersecurity startup Calif found the flaw. Quang Luong used OpenAI's Codex coding assistant — powered by the GPT-6 Cyber model — to audit OpenJDK's internals, and surfaced the bug in about an hour. A first working exploit followed within two hours, at a total AI computing cost of under $200. The New York Times independently verified the vulnerability using proof-of-concept details supplied by Calif.

The bug sits in hand-written processor instructions inside the x86-64 HotSpot Java Virtual Machine, which speed up Base64 decoding on chips with AVX-512 vector extensions. Just 64 "A" characters followed by a single equals sign can make the decoder overwrite Java's memory, leaking sensitive data or letting an attacker run their own code. It affects OpenJDK 17.0.6 and later, including Java 21, 25 and 27, and distributions built from the same code, including Oracle Java SE, Amazon Corretto, Eclipse Temurin, Microsoft Build of OpenJDK, Azul Zulu and Red Hat OpenJDK. Only x86-64 systems with AVX-512 VBMI support are affected; ARM64 systems are not.

Illustration: a cracked digital shield over lines of code. (Generated illustration)

Calif demonstrated unauthenticated remote code execution against widely used applications running in their default configurations: Keycloak, Atlassian Jira and Confluence, JFrog Artifactory, Jenkins and PingFederate. Authenticated attacks were shown against Apache Kafka Broker, Apache Solr, Elasticsearch, Gerrit and Spinnaker. There is no universal exploit — each application needs its own tailored attack — and Oracle rated the flaw difficult to exploit, scoring it 8.1 under CVSS 3.1, while Calif scored it 9.2 out of 10 under CVSS 4.0, the highest "Critical" rating.

Oracle released patches on 17 November, and the US Cybersecurity and Infrastructure Security Agency urged organisations to patch immediately. India's CERT-In and Canada's Cyber Centre issued similar advisories, and South Korea's science ministry urged companies to act. Google said it "swiftly rolled out mitigations and patches across Google services" once notified. As of 17 November there was no evidence the flaw had been exploited in the wild, and it was not on CISA's Known Exploited Vulnerabilities catalogue — but the Canadian centre warned that can change quickly.

A hooded figure at a workstation in a server room. Researchers warn AI is making it faster to find — and exploit — zero-day flaws. (Mental Rogue Studios)

Calif said its research "shows how AI can assist expert security researchers with finding serious flaws in software we all depend on". Fouad Matin, who leads cyber preparedness and safety at OpenAI, said: "We want more security teams to have these capabilities and tools to help make critical software more secure for everyone." The same firm found flaws in PHP in October, and a second firm, Wiz, said its Atlas AI system produced a working proof-of-concept exploit within two hours. Anyone who cannot patch immediately can launch OpenJDK with -XX:-UseBASE64Intrinsics as a temporary mitigation — but every agency's advice is the same: update now.

Sources

  • The Indian Express — "Why a new security flaw in Java has triggered urgent security advisories worldwide" (18 Nov 2026)
  • BleepingComputer — "Critical Java4Shell flaw allows RCE attacks on major Java applications" (17 Nov 2026)
  • DevOps.com — "Java4Shell Vulnerability Threatens Enterprise Java Applications" (18 Nov 2026)
  • Silicon Report — "Critical Java flaw could let attackers take over servers" (17 Nov 2026)
  • Digit — "Java vulnerability that can put millions of systems at risk discovered using AI, US issues warning" (18 Nov 2026)
  • Pivot News — "Java4Shell exploits took hours with AI help, Calif and Wiz…" (18 Nov 2026)
  • MK (Maeil Business Newspaper) — "After a serious security vulnerability was recently discovered in PHP…" (18 Nov 2026)
Advertisement