Researchers Expose Critical Java Flaw with AI Assistance Researchers uncovered CVE-2026-114938, a memory-corruption flaw in OpenJDK's Base64 decoder that enables unauthenticated remote code execution and carries a severity score of 9.2 out of 10, using a human-AI hybrid approach the researchers call a "centaur hacker." The bug could let attackers run commands on a server without credentials. Meet the "centaur hacker" - a powerful hybrid of human and AI that just helped uncover a critical Java flaw, CVE-2026-114938, capable of remote code execution with a near-perfect severity score of 9.2 out of 10. This memory-corruption bug in OpenJDK's Base64 decoder could let attackers run commands on a server without credentials.