What we know so far
A critical memory corruption flaw in OpenJDK's Base64 decoder, tracked as CVE-2026-114938 and nicknamed Java4Shell, lets an unauthenticated attacker overwrite Java heap memory, read secrets from it and reach remote code execution. Fixes shipped last night in Oracle's November Critical Security Patch Update; other Java vendors are still rolling out.
Severity
CVSS 9.2
v4.0 (Calif, OpenJDK). Oracle rates it 8.1 on v3.1.
Impact
Pre-auth RCE
Plus heap memory disclosure
| CVE | CVE-2026-114938 |
| Component | OpenJDK HotSpot x86-64 Base64 decoder intrinsic |
| Bug class | Heap out-of-bounds write (62 bytes) |
| Affected | OpenJDK 17.0.6 and later (17, 21, 25, 27) on x86-64 CPUs with AVX-512 VBMI |
| Not affected | Standard Java 8, Java 11, 17.0.0 to 17.0.5, ARM64 |
| Authentication | None required |
| Exploitation | Exploit attempts seen in the wild (GreyNoise); not yet in CISA KEV |
| Fix | Oracle November 2026 CSPU; other vendors rolling out |
How it works: hand-written assembly that speeds up Base64 decoding using AVX-512 miscalculates the padding on a short, malformed string ending in an equals sign. The decoder then writes a full 64-byte block, 62 bytes past the end of its output buffer, before Java rejects the input. By then the neighbouring heap objects are already corrupted. Calif chained this into a Heartbleed-style memory read it calls JavaBleed, which can leak keys, tokens and personal data before code execution is even attempted.
The flaw has been present in every affected Java release since early 2023. No universal exploit exists: each product needs its own chain, and failed attempts tend to crash the JVM. We will keep this page updated through the day.