Skip to content
Theme
OnDevice Tools

GadgetsLaptops & PCs

Java flaw found by AI can expose critical computer systems

Researchers found a Java vulnerability with help from an AI model, and it could allow attackers to take control of important systems.

Java security warning on a computer screen beside a patch update notice

Key takeaways

  • Calif found the Java flaw with help from OpenAI's GPT-6 Cyber model.
  • The flaw can cause a memory error in Java's base64 decoder.
  • Oracle released a critical patch, and CISA urged immediate fixes.
  • Calif expects some systems may remain without the fix for five to 10 years.
In this article
  1. AI-assisted discovery
  2. How the flaw works
  3. Who is affected
  4. Patch response
  5. Lingering risk and context

AI-assisted discovery

On 18 November 2026, Engadget reported that researchers at cybersecurity startup Calif found a Java flaw with help from OpenAI's GPT-6 Cyber model. The model is not publicly available.

The flaw is in Java, a programming language used by many organisations. Engadget says the bug could affect systems that support banks, hospitals and many large companies.

How the flaw works

On some machines, sending Java's base64 decoder a string of 64 letter A characters followed by an equals sign causes a memory error. A base64 decoder is a component that turns encoded text back into data.

The error usually leads to a crash. Calif says the flaw can be refined to gain complete control without permission.

Who is affected

Engadget reports the bug could allow attackers to compromise systems at banks and hospitals. It could also affect most Fortune 500 companies. Java runs many systems used by banks, hospitals, airlines and government agencies.

Java is used by about 90 percent of the companies on the Fortune 500 list. Calif showed the flaw could be used to attack Keycloak, a login system that helps protect organisations such as CERN and NASA.

Patch response

Illustration: security, Oracle logo, update, settings and the labels Cisa, Java, Oracle

Oracle released a critical patch update on Tuesday. A patch is a software fix. The update did not specifically mention this bug.

The Cybersecurity and Infrastructure Security Agency, known as CISA, instructed organisations to apply fixes without delay. Calif says the flaw is not simple to turn into a practical attack, which may give defenders some time. Organisations should consult official guidance from CISA and Oracle.

Lingering risk and context

Some systems may remain without the fix. Calif expects the bug could remain for five to 10 years.

The report follows recent outages at Microsoft, the FAA, banks, Apple and other services. Engadget says officials have not connected those outages to AI in public statements. The discovery highlights pressure on ageing digital infrastructure.

Sources

  1. AI found a hole in Java, and it starts with 'AAAAAAAA...'Engadget

Tags

Spotted a mistake? See our corrections policy. Read how News by OnDevice works in our editorial policy.