Cybersecurity research firm Calif published details of a potentially major Java security flaw, dubbed Java4Shell, describing an unauthenticated attacker's ability to read Java heap memory, steal private keys, tokens and session secrets, and reach remote code execution. The flaw, tracked as CVE-2026-114938, is a vulnerability in the OpenJDK x86-64 Base64 decoder that scores 9.2 (Critical) under CVSS 4.0, while Oracle rated it 8.1 (High) under CVSS 3.1.
Calif explained that Base64 is a common way of representing binary data as text, used in authentication protocols, file uploads and network messages. OpenJDK uses specialised x86-64 CPU instructions to decode it faster, and the firm found a bug in that optimised machine-level code that lets malformed input overwrite neighbouring memory. The corruption can be triggered by a string of 64 "A" characters followed by an equals sign, or 128 characters in older OpenJDK 17 versions, and with careful exploitation can reveal sensitive memory as Heartbleed did or lead to remote code execution.S1S2
Calif noted that memory corruption is the most common class of vulnerability on the internet, and that Java's memory safety is a major selling point, though much of Java itself is written in low-level languages such as C, C++ and assembly. Assessing the risk is complicated because triggering the bug requires a hardware-accelerated decoding function used on some x86-64 processors supporting AVX-512 VBMI instructions. Fixing it requires patches from a lengthy list of vendors, and installing Tuesday's Oracle patch alone will not resolve it.S1S2
Noah Kenney, principal consultant at Digital 520, said the potential damage from an unpatched exposed Java instance is extreme. He noted that because researchers could read Java heap memory before achieving code execution, an attacker may steal tokens, credentials or keys, and that this access can remain useful even after the vulnerable server is patched.S1S2