The tech start-up Calif used OpenAI's GPT-6 Cyber model to discover a serious security vulnerability in Java's memory functions. The flaw, dubbed Java4Shell, allows an adversary to remotely execute unwanted commands within the system's memory. Because Java is used by 90% of Fortune 500 companies and underpins critical infrastructure in banking, aviation, and telecommunications, experts warned the bug could compromise financial systems and organizational data.
Professor Toby Walsh, a computer scientist and AI researcher, told the BBC that the vulnerability is particularly significant because Java is designed to manage and protect memory for users. "This is why this is such an unlikely and serious flaw," Walsh said.
Oracle, the owner of Java, released a critical patch update on Tuesday containing security vulnerability fixes, though the company has not officially acknowledged the specific Java4Shell bug. It remains unclear if any companies or platforms have been affected by the flaw. Dr. Pedram Hayati, a cyber-security expert, noted that the language is used by almost every bank in Australia alone to manage critical transaction data.
While the vulnerability was identified using a model not yet available to the public, security experts expressed concern that hackers could identify the bug if the flaw is not patched. The discovery follows a recent investigation by Calif into the PHP programming language, which revealed a flaw that had remained undetected for 12 years and affected approximately 30 million websites.














