Calif researchers use OpenAI's GPT-6 Cyber model to find critical Java flaw
Cybersecurity startup Calif discovered a memory-corruption bug in Java's base64 decoder, triggered by a string of 64 'A' characters, that can grant attackers full unauthorized system access. The company demonstrated the exploit against Keycloak, a login system used by organizations including CERN and NASA. Oracle issued a patch this week, and CISA has urged organizations to update immediately.
GoKawiil's interpretation of the reporting above, not reported fact.
Because Java underpins roughly 90% of Fortune 500 systems, an unpatched vulnerability of this kind could leave banks, hospitals and government agencies exposed for years, according to Calif's own estimate of five to ten years of potential lingering risk. The discovery also signals that AI tools are now capable of finding serious vulnerabilities in widely used software, which could accelerate both offensive and defensive cybersecurity efforts. Coming amid a string of unrelated infrastructure outages, it raises broader questions about the resilience of aging systems as AI-assisted security research matures.
- A Java base64 decoder bug can be exploited to gain full unauthorized system access.
- OpenAI's unreleased GPT-6 Cyber model helped Calif researchers identify the flaw.
- Oracle has patched the issue, but Calif warns it could persist in unpatched systems for 5-10 years.
Source: engadget.com — Will Shanklin, 2026-11-18
Published there as: “AI found a hole in Java, and it starts with 'AAAAAAAA...'”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.