A vulnpocalypse bug worthy of attention: Researchers disclose critical flaw in Java with help of AI

Wait 5 sec.

Security researchers on Tuesday published details of a critical flaw in Java, one of the world’s most used programming languages, that could let an attacker take over servers without a password. The group behind the find, Calif Research, said AI coding tools did much of the work. Oracle Corp. shipped a fix the same day it released a November update covering 652 flaws across its product line.The flaw is tracked as CVE-2026-114938 and nicknamed Java4Shell. It is a memory-corruption bug: code that writes data where it does not belong, which can crash a program or, at worst, let an attacker run their own commands. The buggy component is the Base64 decoder in OpenJDK, the free Java runtime that underpins software across banking, aviation and telecommunications, and that, by Calif’s count, runs at 90% of Fortune 500 companies. Calif rated the flaw 9.2 out of 10 on the CVSS scale, where anything above 9 counts as critical. The group said an unauthenticated attacker, one who needs no username or password, could corrupt a server’s memory, read nearby data in a manner similar to the 2014 Heartbleed bug, and escalate that access into full control.The bug lives in an unusual place: hand-written processor instructions that OpenJDK uses to make Base64 decoding faster. The optimization runs only on chips that support a feature set called AVX-512; ARM64 systems are unaffected. Calif found the decoder breaks on a specific input: 64 ‘A’ characters followed by a single equal sign, the padding mark Base64 uses to round out incomplete data. The code then miscalculates a length and writes 64 bytes into a buffer, a small block of memory meant to hold as few as 50, overwriting whatever sits next to it. The program rejects the bad input afterward, but the surrounding memory is already damaged.On its own, that corruption usually crashes the program. Calif showed that an attacker who carefully arranges what sits in the surrounding memory, a technique known as heap grooming, can turn the corruption into control of the server. The group reported unauthenticated code execution in widely used products including Keycloak, Jira, Confluence, Jenkins and PingFederate, and demonstrated a working break-in against Keycloak, a widely used open-source login system. Authenticated execution, requiring a valid account, was shown in Kafka, Solr and Elasticsearch. Researcher samczsun showed Java programs can also be compromised simply by connecting to a malicious server.OpenJDK published a security advisory Tuesday, and Oracle shipped patched Java builds Tuesday afternoon. The find adds to a fast-growing catalog of vulnerabilities surfaced with AI. Calif said researcher Quang Luong found the bug in about an hour auditing OpenJDK with OpenAI’s Codex, for under $200 in computing tokens, and that AI agents produced most of the group’s exploit chains nearly on their own. It was the group’s second major AI-assisted find in recent weeks, after last month’s disclosure of php2shell. In its write-up, the group called this the era of “the centaur hacker, part human, part AI,” and said the hunt for the bug felt like “looking for copper and found gold.”Calif also framed the bug as a byproduct of software’s shared complexity: the AVX-512 speedup shipped with every copy of Java though few programs need it, so an optimization became a critical risk for all. The group said there is no universal exploit and each target requires tailored steps, and it is withholding product-specific exploit details until users have had time to upgrade. No exploitation in the wild had been reported as of this article’s publication, and none of the CVEs related to the issue have been added to CISA’s Known Exploited Vulnerability list.The post A vulnpocalypse bug worthy of attention: Researchers disclose critical flaw in Java with help of AI appeared first on CyberScoop.