Calif says AI found and helped weaponize Java4Shell, a critical remote code execution flaw in OpenJDK

Calif · Company report

Read the source (opens in a new tab)
What happened
Security firm Calif disclosed Java4Shell (CVE-2026-114938), a memory-corruption flaw in OpenJDK's Base64 decoder, on November 17, 2026, the day Oracle released a patch.
Key points
  • Calif says it "used AI to find and weaponize Java4Shell": a researcher found the flaw in an hour and built a first working exploit in two more, for under $200 in tokens.
  • The write-up says the researcher directed Codex to audit OpenJDK, and that later exploit chains for specific applications were built "almost autonomously with AI."
  • Calif's tests showed remote code execution without credentials in default setups of widely used applications including Keycloak, Jira, Confluence and Jenkins; the flaw scores 9.2 of 10 under CVSS 4.0.
  • Calif is holding back its application-specific exploits until users have had a chance to upgrade.
Evidence
The discoverer's own disclosure write-up, published with Oracle's patch. Its time and cost figures are self-reported; it names Codex but no model version.
Uncertain
Calif says the flaw needs an affected OpenJDK build on an x86-64 processor with AVX-512 VBMI support; ARM64 CPUs are not affected.

See it in the latest record