Skip to content
[ SIGNAL VERIFIED ] INDEPENDENT CYBERCRIME JOURNALISMCLASSIFIED // THE UNDERGROUND HAS A NEWSROOM / RSS
.COM
Latest News
ALL NEWS
Newsroom

Broadcom Patches VMware vCenter Zero-Day Exploited in ARMORY CASTLE Espionage

Broadcom fixes CVE-2026-123890, a CVSS 9.8 vCenter authentication bypass. Google says ARMORY CASTLE exploited the zero-day before disclosure.

Broadcom Patches VMware vCenter Zero-Day Exploited in ARMORY CASTLE Espionage
PUBLISHED

IPCRIME | NOVEMBER 19, 2026

Broadcom issued critical VMware security advisory VMSA-2026-0009 on November 19 after Google Threat Intelligence Group and Mandiant linked an authentication-bypass vulnerability in vCenter Server to real-world espionage activity. The main flaw, CVE-2026-123890, carries a CVSS 9.8 rating and allows an attacker with network access to vCenter to gain unauthorized administrative access under affected conditions.

An unusual attack on the virtualization control plane

The issue affects VMware Directory Service authentication. Google says the actor it tracks as ARMORY CASTLE used a tool called BONESAW to exploit the vulnerability, bypass sign-on and create a vSphere administrative account. Its forensic investigation observed use in early September 2026, before the flaw was disclosed publicly. Google assesses the threat group as suspected PRC-nexus; that attribution is a research assessment rather than a court finding.

Broadcom’s fixed releases

The vendor’s response matrix lists vCenter 8.0 Update 3m, 9.0.2.0200 and 9.1.1.0200 for relevant product branches, along with VMware Cloud Foundation 5.2.5 and separate telco-platform guidance. Customers running vCenter 7.0 are directed to Broadcom support if they have an extended-support contract. Broadcom lists no workaround for CVE-2026-123890. A separate set of flaws affects SDDC Manager and VMware Tools; administrators should not assume every item has identical exposure.

Why patching may not fully resolve an intrusion

Control over vCenter can expose directory records, identity tokens and administrative access to managed hosts. In the incident described by Mandiant, the attacker moved laterally from previously compromised remote-support infrastructure before reaching VMware systems. Once a vCenter administrative identity is compromised, rebuilding or updating one binary alone does not necessarily invalidate all access gained before the patch.

Immediate defensive priorities

Identify each vCenter instance and its exact branch, restrict access to its management interfaces, apply the vendor-supported patch and review unexpected SSO accounts, directory changes, privileged sessions and token-signing material where exposure is plausible. Preserve evidence and coordinate identity recovery before resetting trusted components. The published incident does not mean every vCenter deployment has been attacked.

What remains uncertain

The scale of exploitation beyond the disclosed case has not been established publicly. The observed early-September activity and disclosure on November 19 are different dates, and the researcher’s attribution should remain attributed to GTIG and Mandiant.

Primary sources and related coverage

Documentation-based IPCRIME reporting; no independent forensic or exploit testing claimed. Featured image is conceptual artwork, not incident footage.

Editorial note

IPCRIME distinguishes public records and documented findings from allegations, analysis and commentary. Report corrections through our contact page.