

A string of 64 “A” characters may look harmless, but under specific conditions it can trigger a serious Java security flaw, potentially allowing hackers to take control of computer systems. Cybersecurity startup Calif uncovered the vulnerability with help from OpenAI’s GPT-6 Cyber model, Engadget reported.
The discovery has prompted warnings to organizations that rely on Java, used in banking, healthcare and government.
The problem lies in Java’s Base64 decoder, which converts encoded text into data. Researchers found that feeding it 64 “A” characters followed by an equals sign can trigger a memory error.
That error would usually crash a system. Calif said, however, that the issue could be manipulated to enable unauthorized access and remote code execution. An attacker could potentially run malicious instructions on a vulnerable system without logging in.
Calif demonstrated the risk in Keycloak, a login platform used by CERN and NASA. Java is used by around 90 percent of Fortune 500 companies, Engadget reported, making the potential exposure significant.
The vulnerability is not considered easy to exploit, giving defenders some time to respond. However, the US Cybersecurity and Infrastructure Security Agency has urged organizations to patch affected systems immediately. Oracle issued a critical Java update on November 17, although it did not specifically identify this flaw, Engadget reported.
US, South Korean, German and British authorities, along with China’s Tencent, also warned organizations about the vulnerability, known as Java4Shell, Reuters reported. The Dutch cybersecurity agency said it was investigating. German and Dutch authorities reported no known successful attacks at the time.
Also Read: Developing Secure Microservices with Java
The discovery highlights how artificial intelligence is increasingly being used to identify weaknesses in widely deployed software. The same capabilities could help attackers find and exploit flaws faster, leaving organizations less time to respond.
Calif estimates that unpatched systems could remain vulnerable for five to 10 years. The warning underlines the need for security teams to identify affected Java installations and apply available updates promptly.