Purnima Padmanabhan, VP and GM, Broadcom's Tanzu Division
– Purnima Padmanabhan, VP/GM & Kevin Strohmeyer, marketing head, Broadcom Tanzu (Giacomo Lee/SDxCentral)

LONDON – Broadcom warned to expect more dramatic cybersecurity vulnerabilities following this week’s discovery of the major Java4Shell flaw in programming bedrock Java.

The remarks came at this week’s European edition of VMware Explore, where Purnima Padmanabhan, VP and GM of VMware’s Tanzu division, commented the firm was tracking the flaw, with a workaround already deployed and patches on the way for users of Tanzu’s flagship platform and application engineering services.

Java4Shell, which was discovered by the same researchers who uncovered the PHP2Shell vulnerability, is a critical memory-corruption issue which allows unauthenticated access to the main pool of runtime memory in Java Virtual Machine (JVM) that allows for remote code execution (RCE). Various worldwide security bodies, including the Cybersecurity and Infrastructure Security Agency (CISA), issued urgent advisories, while Oracle and Amazon shipped emergency updates to prevent breaches.

Unlike with PHPShell, no organizations have publicly confirmed an attack relating to the flaw, though its finding is the latest in a long line of recent technology mishaps to befall infrastructure in recent weeks. SDxCentral understood that Broadcom customer Deutsche Bank had to pull out from speaking at the event due to a security breach on the same day.

Broadcom on frontier AI-driven cybersecurity

Commenting on Java4Shell’s discovery using OpenAI’s GPT-6 Cyber model, Padmanabhan said the deterministic nature of such models means platform-wide scans are not possible, with the discovery process a continuous one.

“There's still something else that could be equally dramatic that could be found very soon,” Padmanabhan said. “Don't expect this is cleaned up. We are going to find very dramatic things.”

Padmanabhan also revealed the firm is using a mix of low-cost and high-cost models in its vulnerability research, using Claude Mythos as part of its ongoing membership of Anthropic’s Project Glasswing initiative, as well as various GPT models from OpenAI, not just limited to GPT-6 Cyber.

The VP also reminded media of Tanzu’s maintainer status of the open-source Spring Java framework.

“Even before Mythos, because we are the maintainers of Spring, all security advisories come to us," Padmanabhan explained. "We had seen a 2,000% increase in security findings. Actually, 1,700-plus percentage. So, while Mythos is just absolutely dramatic and insane, I mean, crazy what we find with it, we are finding it with all models, even the Daybreak and OpenAI stuff.”