Morning Edition · №
Cybersecurity SAN FRANCISCO, CALIF.

AI Uncovers a Decade-Old Critical Flaw at the Heart of Java, One of the World's Most Used Languages

A security startup used an unreleased OpenAI model to find a memory-handling bug in Java that has gone undetected for more than ten years, arriving amid a run of major outages across banks, airlines and government agencies.

SHARE X f in ⧉

A cybersecurity startup says it has used artificial intelligence to uncover a critical, previously unknown flaw in Java, the programming language that underpins much of the world's banking, government and telecommunications infrastructure — the second such discovery in as many months after the same company found a comparable bug in PHP.

The firm, Calif, found the vulnerability in functions Java uses to manage computer memory, and named it Java4Shell. Java is used by nearly all of the world's largest companies, including Apple and Amazon, as well as banks, hospitals and government agencies; one of its core selling points has long been that it manages memory automatically, in theory making it harder to exploit than languages that leave memory handling to programmers. "This is such an unlikely and serious flaw," said Toby Walsh, a computer scientist who reviewed the findings, of a bug sitting precisely in the part of the language designed to prevent this category of problem.

Found by one AI model, exploited in testing by another

Calif said it discovered Java4Shell with the help of AI, and that the system used to actually exploit the flaw in testing was GPT-6 Cyber, an OpenAI model not yet available to the public. The vulnerability allows an attacker to remotely execute unwanted commands within Java's memory system, a class of flaw that cybersecurity specialists regard as among the most dangerous because overloading or corrupting a program's memory is one of the most common ways intruders exploit or crash a system. Oracle, which owns Java, released a "critical patch update" on Tuesday that it said fixed the flaw among other issues; it was not immediately clear whether any companies or platforms had been compromised before the patch was issued.

Dr Pedram Hayati, a cybersecurity researcher, said the flaw could compromise "critical financial systems, customers and organisation's data," noting that in Australia alone almost every bank relies on Java to manage transaction records. Java has been in use since 1995 and, in Calif's own description of its investigation, now functions as "the beating heart of the modern economy."

Arriving amid a string of unrelated outages

The discovery lands in the middle of an unusually dense run of technology failures. A nationwide ground stop halted US flights for several hours on November 13 after a router malfunction at an FAA communications facility; Apple's iCloud, App Store and account sign-in services went down on November 17 because of what the company called an internal update unrelated to any attack; and the National Security Agency's internal network went offline this week for what it described as scheduled maintenance. Secretary of State Marco Rubio said there was "zero evidence" the incidents were connected, and officials have not linked any of them to the newly discovered Java flaw. Earlier in November, OpenAI president Greg Brockman said the company was shifting a quarter of its staff onto cybersecurity work, citing the pace at which AI tools are now surfacing vulnerabilities that had sat undetected for years.

The PHP flaw that Calif identified in October, by comparison, had gone undetected for roughly a decade and left an estimated 30 million websites exposed before site operators were urged to patch or take affected systems offline. Walsh said the pattern — flaws "hiding in plain sight" for more than a decade in two of the internet's most foundational languages — reflects AI's growing ability to find classes of bugs that would have been too obscure or labor-intensive for human researchers to locate on their own. "AI is now exposing more flaws much more quickly," he said.

SHARE THIS ARTICLE X Facebook LinkedIn Copy link
Claire Fontaine · Technology & Regulation Correspondent

Reports on technology and its regulation for UBStandard, with a focus on Brussels, AI policy and Europe's digital economy.

[email protected]
Related coverage Front page →