A cybersecurity startup used artificial intelligence to uncover a critical vulnerability in Java, the programming language that underpins banking, aviation and national-security systems worldwide.
The flaw, dubbed Java4Shell, sits inside the memory-management functions Java was designed to protect. Researchers at Calif found it with help from OpenAI’s recently released model GPT-6 Cyber.
Java flaw threatens core systems
Java runs on 90% of Fortune 500 companies and nearly every major bank, hospital and government agency. First released in 1995, it has become the “beating heart of the modern economy,” according to Calif’s investigation. One core promise is automatic memory management — a safeguard meant to prevent the very class of exploit now exposed.
“This is why this is such an unlikely and serious flaw,” said Professor Toby Walsh, a computer scientist and one of the world’s leading AI researchers, speaking to the BBC. “Java is meant to make the IT world more secure by managing your memory online.”
The vulnerability lets an adversary remotely execute unwanted commands against Java’s memory system. Cybersecurity expert Dr Pedram Hayati warned it could compromise “critical financial systems, customers and organisation’s data.” In Australia alone, he said, almost every bank relies on Java for managing critical transaction data.
AI finds bugs humans missed
Calif’s use of GPT-6 Cyber illustrates a new dynamic: AI models are finding flaws that human researchers missed for over a decade. “It’s been hiding in plain sight for 12 years,” Walsh said. “AI is now exposing more flaws much more quickly,” he added, including ones too obscure for humans to detect alone.
The discovery follows Calif’s earlier identification of a critical flaw in PHP, the language behind sites such as Wikipedia. That vulnerability affected an estimated 30 million websites, according to Professor Alan Woodward of the Surrey Centre for Cyber Security. The PHP flaw had gone undetected for at least a decade — a pattern Walsh sees repeating with Java.
Oracle silent on specific bug
Oracle, which owns Java, has not publicly acknowledged the specific bug. On Tuesday it released a “critical patch update” containing security vulnerability fixes, though it remains unclear whether the Java4Shell flaw is addressed. No platforms or companies have confirmed being affected by the vulnerability.
The NSA did not specify whether its outage was connected to the Java bug. Oracle’s patch timeline and the extent of any exploitation remain the critical unknowns as administrators weigh emergency updates against operational risk.